This study was supported by flora robotica project that received funding from the European Union&#39;s Horizon 2020 research and innovation programme under the FET grant agreement, no. 640959. The authors thank Anastasios Getsopulos and Ewald Neufeld for their contribution in hardware assembly, and Tanja Katharina Kaiser for her contribution in monitoring plant experiments.

<ol>
	<li>&#197;strand, B., Baerveldt, A. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=An+agricultural+mobile+robot+with+vision-based+perception+for+mechanical+weed+control.">An agricultural mobile robot with vision-based perception for mechanical weed control.</a> Autonomous Robots. 13 (1), 21-35 (2002).</li><li>Blackmore, B. S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+systems+view+of+agricultural+robots.">A systems view of agricultural robots.</a> Proceedings of 6th European conference on precision agriculture (ECPA). , 23-31 (2007).</li><li>Edan, Y., Han, S., Kondo, N. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Automation+in+agriculture.">Automation in agriculture.</a> Springer handbook of automation. , 1095-1128 (2009).</li><li>Van Henten, E. J., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=An+autonomous+robot+for+harvesting+cucumbers+in+greenhouses.">An autonomous robot for harvesting cucumbers in greenhouses.</a> Autonomous Robots. 13 (3), 241-258 (2002).</li><li>Al-Beeshi, B., Al-Mesbah, B., Al-Dosari, S., El-Abd, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=iplant:+The+greenhouse+robot.">iplant: The greenhouse robot.</a> Proceedings of IEEE 28th Canadian Conference on Electrical and Computer Engineering (CCECE). , 1489-1494 (2015).</li><li>Giraldo, J. P., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Plant+nanobionics+approach+to+augment+photosynthesis+and+biochemical+sensing.">Plant nanobionics approach to augment photosynthesis and biochemical sensing.</a> Nature Materials. 13 (4), (2014).</li><li>Mazarei, M., Teplova, I., Hajimorad, M. R., Stewart, C. N. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Pathogen+phytosensing:+Plants+to+report+plant+pathogens.">Pathogen phytosensing: Plants to report plant pathogens.</a> Sensors. 8 (4), 2628-2641 (2008).</li><li>Zimmermann, M. R., Mith&#246;fer, A., Will, T., Felle, H. H., Furch, A. C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Herbivore-triggered+electrophysiological+reactions:+candidates+for+systemic+signals+in+higher+plants+and+the+challenge+of+their+identification.">Herbivore-triggered electrophysiological reactions: candidates for systemic signals in higher plants and the challenge of their identification.</a> Plant Physiology. , 01736 (2016).</li><li>Hamann, H., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> Flora robotica--An Architectural System Combining Living Natural Plants and Distributed Robots. , (2017).</li><li>Arkin, R. C., Egerstedt, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Temporal+heterogeneity+and+the+value+of+slowness+in+robotic+systems.">Temporal heterogeneity and the value of slowness in robotic systems.</a> Proceedings of IEEE International Conference on Robotics and Biomimetics (ROBIO). , 1000-1005 (2015).</li><li>Mahlein, A. K. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Plant+disease+detection+by+imaging+sensors-parallels+and+specific+demands+for+precision+agriculture+and+plant+phenotyping).">Plant disease detection by imaging sensors-parallels and specific demands for precision agriculture and plant phenotyping).</a> Plant Disease. 100 (2), 241-251 (2016).</li><li>Wahby, M., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+robot+to+shape+your+natural+plant:+the+machine+learning+approach+to+model+and+control+bio-hybrid+systems.">A robot to shape your natural plant: the machine learning approach to model and control bio-hybrid systems.</a> Proceedings of the Genetic and Evolutionary Computation Conference (GECCO &#39;18). , 165-172 (2018).</li><li>Bastien, R., Douady, S., Moulia, B. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+unified+model+of+shoot+tropism+in+plants:+photo-,+gravi-and+propio-ception.">A unified model of shoot tropism in plants: photo-, gravi-and propio-ception.</a> PLoS Computational Biology. 11 (2), e1004037 (2015).</li><li>Wahby, M., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Autonomously+shaping+natural+climbing+plants:+a+bio-hybrid+approach.">Autonomously shaping natural climbing plants: a bio-hybrid approach.</a> Royal Society Open Science. 5 (10), 180296 (2018).</li><li>Liscum, E., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Phototropism:+growing+towards+an+understanding+of+plant+movement.">Phototropism: growing towards an understanding of plant movement.</a> Plant Cell. 26, 38-55 (2014).</li><li>Christie, J. M., Murphy, A. S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Shoot+phototropism+in+higher+plants:+new+light+through+old+concepts.">Shoot phototropism in higher plants: new light through old concepts.</a> American Journal of Botany. 100, 35-46 (2013).</li><li>Migliaccio, F., Tassone, P., Fortunati, A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Circumnutation+as+an+autonomous+root+movement+in+plants.">Circumnutation as an autonomous root movement in plants.</a> American Journal of Botany. 100, 4-13 (2013).</li><li>Gianoli, E. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+behavioural+ecology+of+climbing+plants.">The behavioural ecology of climbing plants.</a> AoB Plants. 7, (2015).</li><li>Vestartas, P., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Design+Tools+and+Workflows+for+Braided+Structures.">Design Tools and Workflows for Braided Structures.</a> Proceedings of Humanizing Digital Reality. , 671-681 (2018).</li><li>Pierik, R., De Wit, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Shade+avoidance:+phytochrome+signalling+and+other+aboveground+neighbour+detection+cues.">Shade avoidance: phytochrome signalling and other aboveground neighbour detection cues.</a> Journal of Experimental Botany. 65 (10), 2815-2824 (2014).</li><li>Fraser, D. P., Hayes, S., Franklin, K. A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Photoreceptor+crosstalk+in+shade+avoidance.">Photoreceptor crosstalk in shade avoidance.</a> Current Opinion in Plant Biology. 33, 1-7 (2016).</li><li>Hogewoning, S. W., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Photosynthetic+Quantum+Yield+Dynamics:+From+Photosystems+to+Leaves.">Photosynthetic Quantum Yield Dynamics: From Photosystems to Leaves.</a> The Plant Cell. 24 (5), 1921-1935 (2012).</li><li>McCree, K. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+action+spectrum,+absorptance+and+quantum+yield+of+photosynthesis+in+crop+plants.">The action spectrum, absorptance and quantum yield of photosynthesis in crop plants.</a> Agricultural Meteorology. 9, 191-216 (1971).</li><li>. Autonomously shaping natural climbing plants: a bio-hybrid approach [Dataset] Available from: <a target="_blank" href="https://doi.org/10.5281/zenodo.1172160">https://doi.org/10.5281/zenodo.1172160</a> (2018)</li></ol>

The authors declare that they have no competing financial interests.

The presented methodology shows initial steps toward automating the stimuli-driven steering of plant growth, to generate specific patterns. This requires continuous maintenance of plant health while combining into a single experiment setup the distinct realms of biochemical growth responses and engineered mechatronic functions-sensing, communication, and controlled generation of stimuli. As our focus here is on climbing plants, mechanical support is also integral. A limitation of the current setup is its scale, but we believe our methodology easily scales. The mechanical scaffold can be extended for larger setups and therefore longer periods of growth, which also allows expanded configurations and patterns. Here the setup is limited to two dimensions and binary left-right decisions, as growth is limited to a grid of mechanical supports at 45&#176; inclination, and plant decision positions are limited to that grid&#39;s bifurcations. Mechanical extensions may include 3D scaffolds and differing materials, to allow for complex shapes9,19. The methodology can be considered a system to automatically grow patterns defined by a user. By extending the possible complexity of mechanical configurations, users should face few restrictions on their desired patterns. For such an application, a user software tool should confirm that the pattern is producible, and the mechatronics should then self-organize the production of the pattern by generating appropriate stimuli to steer the plants. The software should also be extended to include recovery plans and policies determining how to continue with the growth if the original planned pattern has partially failed-for instance if the first activated robot has never detected a plant but the dormant ones have seen that the position of the growing tips are beyond the activated robot.
In the presented methodology, an example plant species meeting the protocol selection criteria is the climbing common bean, P. vulgaris. This is the species used in the representative results. As P. vulgaris has strong positive phototropism to UV-A and blue light, the phototropins (light-receptor proteins) in the plant will absorb photons corresponding to wavelengths 340-500 nm. When the receptors are triggered, first swelling will occur in the stem by the preferential relocation of water to the stem tissues opposing the triggered receptors, causing a reversible directional response. Then, within the stem, auxin (plant patterning hormone) is directed to the same tissue location, perpetuating the directional response and fixing stem tissues as they stiffen. This behavior can be used for shaping the plants in these controlled indoor conditions, as the plants are exposed only to isolated blue light and isolated red light, with incident far-red light from IR-proximity sensors at low enough levels that it does not interfere with behaviors such as the shade-avoidance response20,21. The phototropism reaction in the plant responds in the setup to light from blue diodes with peak emission &#411;max = 465 nm, and photosynthesis22,23 in the plant is supported by red diodes with peak emission &#411;max = 650 nm. P. vulgaris growing up to several meters in height is suitable in the overall setup, as the roughly 3 L of commercial gardening soil needed per pot fits the setup scale.
Although the current setup focuses on light as an attraction stimulus, additional stimuli may be relevant for other experiment types. If the desired pattern requires a separation between different groups of plants (e.g., the desired pattern needs two groups of plants to choose opposite sides), then it may not be feasible using only one type of stimulus. For such complex growth patterns independent of scaffold shape, the different groups of plants can potentially be grown in different time periods such that their respective attraction stimuli do not interfere, which would also allow the integration of branching events. However, this may not always be a suitable solution, and the standard attractive light stimulus could then be augmented by repelling influences such as shading, or by other stimuli like far-red light or vibration motors9,14.
The presented method and the experiment design are only an initial first step towards a sophisticated methodology to automatically influence directional growth of plants. The experiment setup is basic by determining only a sequence of binary decisions in the plants and we focus on one, easy to manage stimulus. Additional studies would be required to prove the method&#39;s statistical significance, to add more stimuli, and to control other processes such as branching. With sufficient development to guarantee the long-term reliability of the robots, the presented methodology could allow for automation of plant experiments over long time periods, reducing the overhead associated with the study of plant development stages beyond that of shoots. Similar methods can allow for future investigations into the underexplored dynamics between biological organisms and autonomous robots, when the two act as tightly coupled self-organizing bio-hybrid systems.

Congruent with the increasing prevalence of automation in manufacturing and production, robots are being utilized to sow, treat, and harvest plants1,2,3,4,5. We use robot technology to automate plant experiments in a non-invasive manner, with the purpose of steering growth via directional responses to stimuli. Traditional gardening practices have included the manual shaping of trees and bushes by mechanical restraint and cutting. We present a methodology that can for instance be applied to this shaping task, by steering growth patterns with stimuli. Our presented methodology is also a step towards automated plant experiments, here with a specific focus on providing light stimuli. Once the technology has become robust and reliable, this approach has potential to reduce costs in plant experiments and to allow for new automated experiments that would otherwise be infeasible due to overhead in time and manual labor. The robotic elements are freely programmable and act autonomously as they are equipped with sensors, actuators for stimuli provision, and microprocessors. While we focus here on proximity sensing (i.e., measuring distances at close-range) and light stimuli, many other options are feasible. For example, sensors can be used to analyze plant color, to monitor biochemical activity6, or for phytosensing7 approaches to monitor for instance environmental conditions through plant electrophysiology8. Similarly, actuator options might provide other types of stimuli9, through vibration motors, spraying devices, heaters, fans, shading devices, or manipulators for directed physical contact. Additional actuation strategies could be implemented to provide slow mobility to the robots (i.e., &#39;slow bots&#39;10), such that they could gradually change the position and direction from which they provide stimuli. Furthermore, as the robots are equipped with single-board computers, they could run more sophisticated processes such as visioning for plant phenotyping11 or artificial neural network controllers for stimuli actuation12. As the plant science research focus is often on early growth (i.e., in shoots)13, the whole domain of using autonomous robot systems to influence plants over longer periods seems underexplored and may offer many future opportunities. Going even one step further, the robotic elements can be seen as objects of research themselves, allowing the study of the complex dynamics of bio-hybrid systems formed by robots and plants closely interacting. The robots selectively impose stimuli on the plants, the plants react according to their adaptive behavior and change their growth pattern, which is subsequently detected by the robots via their sensors. Our approach closes the behavioral feedback loop between the plants and the robots and creates a homeostatic control loop.
In our experiments to test the function of the robot system, we exclusively use the climbing common bean, Phaseolus vulgaris. In this setup, we use climbing plants, with mechanical supports in a gridded scaffold of overall height 180 cm, such that the plants are influenced by thigmotropism and have a limited set of growth directions to choose from. Given that we want to shape the whole plant over a period of weeks, we use blue light stimuli to influence the plant&#39;s phototropism macroscopically, over different growth periods including young shoots and later stem stiffening. We conduct the experiments in fully controlled ambient light conditions where other than the blue light stimuli we provide exclusively red light, with peak emission at wavelength 650 nm. When they reach a bifurcation in the mechanical support grid, they make a binary decision whether to grow left or right. The robots are positioned at these mechanical bifurcations, separated by distances of 40 cm. They autonomously activate and deactivate their blue light emittance, with peak emission at wavelength 465 nm, according to a predefined map of the desired growth pattern (in this case, a zigzag pattern). In this way, the plants are guided from bifurcation to bifurcation in a defined sequence. Only one robot is activated at a given time&#8212;during which it emits blue light while autonomously monitoring plant growth on the mechanical support beneath it. Once it detects a growing tip using its infrared proximity sensors, it stops emitting blue light and communicates to its neighboring robots via radio. The robot that determines itself to be the next target in the sequence then subsequently activates, attracting plant growth toward a new mechanical bifurcation.
As our approach incorporates both engineered and natural mechanisms, our experiments include several methods that operate simultaneously and interdependently. The protocol here is first organized according to the type of method, each of which must be integrated into a unified experiment setup. These types are plant species selection; robot design including hardware and mechanics; robot software for communication and control; and the monitoring and maintenance of plant health. The protocol then proceeds with the experiment design, followed by data collection and recording. For full details of results obtained so far, see Wahby et al.14. Representative results cover three types of experiments&#8212;control experiments where all robots do not provide stimuli (i.e., are dormant); single-decision experiments where the plant makes a binary choice between one stimuli-providing robot and one that is dormant; and multiple-decision experiments where the plant navigates a sequence of binary choices to grow a predefined pattern.

<table>
	<tbody>
		<tr>
			<td>3D printed case</td>
			<td>Shapeways, Inc</td>
			<td>n/a</td>
			<td>Customized product, https://www.shapeways.com/</td>
		</tr>
		<tr>
			<td>3D printed joints</td>
			<td>n/a</td>
			<td>n/a</td>
			<td>Produced by authors</td>
		</tr>
		<tr>
			<td>Adafruit BME280 I2C or SPI Temperature Humidity Pressure Sensor</td>
			<td>Adafruit</td>
			<td>2652</td>
			<td></td>
		</tr>
		<tr>
			<td>Arduino Uno Rev 3</td>
			<td>Arduino</td>
			<td>A000066</td>
			<td></td>
		</tr>
		<tr>
			<td>CdS photoconductive cells</td>
			<td>Lida Optical &#38; Electronic Co., Ltd</td>
			<td>GL5528</td>
			<td></td>
		</tr>
		<tr>
			<td>Cybertronica PCB</td>
			<td>Cybertronica Research</td>
			<td>n/a</td>
			<td>Customized product, http://www.cybertronica.de.com/download/D2_node_module_v01_appNote16.pdf</td>
		</tr>
		<tr>
			<td>DC Brushless Blower Fan</td>
			<td>Sunonwealth Electric Machine Industry Co., Ltd.</td>
			<td>UB5U3-700</td>
			<td></td>
		</tr>
		<tr>
			<td>Digital temperature sensor</td>
			<td>Maxim Integrated</td>
			<td>DS18B20</td>
			<td></td>
		</tr>
		<tr>
			<td>High Power (800 mA) EPILED - Far Red / Infra Red (740-745 nm)</td>
			<td>Future Eden Ltd.</td>
			<td>n/a</td>
			<td></td>
		</tr>
		<tr>
			<td>I2C Soil Moisture Sensor</td>
			<td>Catnip Electronics</td>
			<td>v2.7.5</td>
			<td></td>
		</tr>
		<tr>
			<td>IR-proximity sensors (4-30 cm)</td>
			<td>Sharp Electronics</td>
			<td>GP2Y0A41SK0</td>
			<td></td>
		</tr>
		<tr>
			<td>LED flashlight (50 W)</td>
			<td>Inter-Union Technohandel GmbH</td>
			<td>103J50</td>
			<td></td>
		</tr>
		<tr>
			<td>LED Red Blue Hanging Light for Indoor Plant (45 W)</td>
			<td>Erligpowht</td>
			<td>B00S2DPYQM</td>
			<td></td>
		</tr>
		<tr>
			<td>Low-voltage submersible pump 600 l/h (6 m rise)</td>
			<td>Peter Barwig Wasserversorgung</td>
			<td>444</td>
			<td></td>
		</tr>
		<tr>
			<td>Medium density fibreboard</td>
			<td>n/a</td>
			<td>n/a</td>
			<td>For stand</td>
		</tr>
		<tr>
			<td>Micro-Spectrometer (Hamamatsu) on an Arduino-compatible breakout board</td>
			<td>Pure Engineering LLC</td>
			<td>C12666MA</td>
			<td></td>
		</tr>
		<tr>
			<td>Pixie - 3W Chainable Smart LED Pixel</td>
			<td>Adafruit</td>
			<td>2741</td>
			<td></td>
		</tr>
		<tr>
			<td>Pots (3.5 l holding capacity, 15.5 cm in height)</td>
			<td>n/a</td>
			<td>n/a</td>
			<td></td>
		</tr>
		<tr>
			<td>Power supplies (5 V, 10 A)</td>
			<td>Adafruit</td>
			<td>658</td>
			<td></td>
		</tr>
		<tr>
			<td>Raspberry Pi 3 Model B</td>
			<td>Raspberry Pi Foundation</td>
			<td>3B</td>
			<td></td>
		</tr>
		<tr>
			<td>Raspberry Pi Camera Module V2</td>
			<td>Raspberry Pi Foundation</td>
			<td>V2</td>
			<td></td>
		</tr>
		<tr>
			<td>Raspberry Pi Zero</td>
			<td>Raspberry Pi Foundation</td>
			<td>Zero</td>
			<td></td>
		</tr>
		<tr>
			<td>RGB Color Sensor with IR filter and White LED - TCS34725</td>
			<td>Adafruit</td>
			<td>1334</td>
			<td></td>
		</tr>
		<tr>
			<td>Sowing and herb soil</td>
			<td>Gardol</td>
			<td>n/a</td>
			<td></td>
		</tr>
		<tr>
			<td>String bean</td>
			<td>SPERLI GmbH</td>
			<td>402308</td>
			<td></td>
		</tr>
		<tr>
			<td>Transparent acrylic 5 mm sheet</td>
			<td>n/a</td>
			<td>n/a</td>
			<td>For supplemental structural support</td>
		</tr>
		<tr>
			<td>Wooden rods (birch wood), painted black, 5 mm diameter</td>
			<td>n/a</td>
			<td>n/a</td>
			<td>For plants to climb</td>
		</tr>
	</tbody>
</table>

robotic sensing and stimuli provision for guided plant growth

Robot systems are actively researched for manipulation of natural plants, typically restricted to agricultural automation activities such as harvest, irrigation, and mechanical weed control. Extending this research, we introduce here a novel methodology to manipulate the directional growth of plants via their natural mechanisms for signaling and hormone distribution. An effective methodology of robotic stimuli provision can open up possibilities for new experimentation with later developmental phases in plants, or for new biotechnology applications such as shaping plants for green walls. Interaction with plants presents several robotic challenges, including short-range sensing of small and variable plant organs, and the controlled actuation of plant responses that are impacted by the environment in addition to the provided stimuli. In order to steer plant growth, we develop a group of immobile robots with sensors to detect the proximity of growing tips, and with diodes to provide light stimuli that actuate phototropism. The robots are tested with the climbing common bean, Phaseolus vulgaris, in experiments having durations up to five weeks in a controlled environment. With robots sequentially emitting blue light-peak emission at wavelength 465 nm-plant growth is successfully steered through successive binary decisions along mechanical supports to reach target positions. Growth patterns are tested in a setup up to 180 cm in height, with plant stems grown up to roughly 250 cm in cumulative length over a period of approximately seven weeks. The robots coordinate themselves and operate fully autonomously. They detect approaching plant tips by infrared proximity sensors and communicate via radio to switch between blue light stimuli and dormant status, as required. Overall, the obtained results support the effectiveness of combining robot and plant experiment methodologies, for the study of potentially complex interactions between natural and engineered autonomous systems.

Control: Plant Behavior without Robotic Stimuli. 
Due to the lack of blue light (i.e., all robots are dormant), positive phototropism is not triggered in the plant. Therefore, the plants show unbiased upwards growth as they follow gravitropism. They also display typical circumnutation (i.e., winding), see Figure 4A. As expected, the plants fail to find the mechanical support leading to the dormant robots. The plants collapse when they can no longer support their own weight. We stop the experiments when at least two plants collapse, see Figure 4B,C.
Single or Multiple Decisions: Plant Behavior with Robotic Stimuli 
In four single-decision experiments, two runs have leftward steering (i.e., the robot left of the bifurcation is activated to stimulus), and two runs have rightward steering. The stimulus robots successfully steer the plants towards the correct support, see Figure 5. The nearest plant with stem angle most similar to that of the correct support attaches first. In each experiment, at least one plant attaches to the support and climbs it until it reaches the stimulus robot and thereby ends the experiment. In one experiment, a second plant attaches to the correct support. The remaining plants might attach as well in longer experiment durations. None of the plants attaches to the incorrect support. Each experiment runs continuously for 13 days on average.
In two multiple-decision experiments, the plants grow into a predefined zigzag pattern, see Figure 6A. Each experiment runs for approximately seven weeks. As an experiment starts, a robot sets its status to stimulus (see 3.6.3) and steers the plants towards the correct support according to the stipulated pattern. A plant attaches and climbs it, arriving at the activated stimulus robot therefore completing the first decision. According to 3.7.3, the current stimulus robot then becomes dormant and notifies its adjacent neighbors. The dormant neighbor that is next on the zigzag pattern switches itself to stimulus (see 3.7.6). If a plant is detected by a dormant robot, that robot does not react (see 3.7.2). The plants continue and complete the remaining three decisions successfully. The predefined zigzag pattern is therefore fully grown, see Figure 6B.
All experiment data, as well as videos, are available online24.
<img alt="Figure 1" class="xfigimg" src="/files/ftp_upload/59835/59835fig1.jpg" /> 
Figure 1. The immobile robot and its primary components. Figure reprinted from author publication Wahby et al.14, used with Creative Commons license CC-BY 4.0 (see supplemental files), with modifications as permitted by license. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig1large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 2" class="xfigimg" src="/files/ftp_upload/59835/59835fig2.jpg" /> 
Figure 2. The component diagram of the immobile robot electronics. IRLML2060 LED drivers are interfaced with the robot&#39;s single-board computer (e.g. Raspberry Pi) via PWM to control the brightness of the LEDs. An LP5907 switch is interfaced with the single-board computer via general-purpose input/output (GPIO) header pin, to control the fan. An MCP3008 analog-to-digital converter (ADC) is interfaced with the single-board computer via serial peripheral interface (SPI) to read the analog IR and light-dependent resistor (LDR) sensor data. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 3" class="xfigimg" src="/files/ftp_upload/59835/59835fig3.jpg" /> 
Figure 3. Shortly after &#39;03.04.16,&#39; a plant tip climbs a support and arrives in the field of view of the robot. (A) Sample IR-proximity sensor scaled voltage readings (vertical axis) during an experiment. Higher values indicate plant tip detection. (B) The IR-proximity sensor is placed and oriented according to the support attachment, to ensure effective plant tip detection. Figure reprinted from author publication Wahby et al.14, used with Creative Commons license CC-BY 4.0 (see supplemental files), with modifications as permitted by license. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig3large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 4" class="xfigimg" src="/files/ftp_upload/59835/59835fig4.jpg" /> 
Figure 4. Control experiments result frames showing that all four plants did not attach to any support in the absence of blue light. (A) After five days, all plants growing upwards in one of the control experiments (see (C) for later growth condition). (B) After 15 days, three plants collapsed, and one still growing upwards in the first control experiment. (C) After seven days, two plants collapsed, and two still growing upwards in the second control experiment (see (A) for previous growth condition). Figure reprinted from author publication Wahby et al.14, used with Creative Commons license CC-BY 4.0 (see supplemental files), with modifications as permitted by license. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig4large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 5" class="xfigimg" src="/files/ftp_upload/59835/59835fig5.jpg" /> 
Figure 5. Single-decision experiments result frames showing the ability of a stimulus robot to steer the plants through a binary decision, to climb the correct support. In all four experiments, one robot is set to stimulus and the other to dormant-at two opposite sides of a junction. The frames show the plants&#39; location right before the stimulus robot detects them. In each experiment at least one plant attaches to the correct support, and no plant attaches to the incorrect one. Also, the unsupported plants show growth biased towards the stimulus robot. E, F, G, H are closeups of A, B, C, D respectively. Figure reprinted from author publication Wahby et al.14, used with Creative Commons license CC-BY 4.0 (see supplemental files), with modifications as permitted by license. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig5large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 6" class="xfigimg" src="/files/ftp_upload/59835/59835fig6v2.jpg" /> 
Figure 6. Multiple-decision experiment. (A) The targeted zigzag pattern is highlighted in green on the map. (B) The last frame from the experiment (after 40 days), showing the plants&#39; situation before the last stimulus robot on the pattern detects them. The robots successfully grow the zigzag pattern. Figure reprinted from author publication Wahby et al.14, used with Creative Commons license CC-BY 4.0 (see supplemental files), with modifications as permitted by license. <a href="https://www.jove.com/files/ftp_upload/59835/59835fig6large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>

Watch this Scientific Journal Video about Robotic Sensing and Stimuli Provision for Guided Plant Growth at JoVE.com

Robotic Sensing and Stimuli Provision for Guided Plant Growth

Distributed robot nodes provide sequences of blue light stimuli to steer the growth trajectories of climbing plants. By activating natural phototropism, the robots guide the plants through binary left-right decisions, growing them into predefined patterns that by contrast are not possible when the robots are dormant.

Robot systems are actively researched for manipulation of natural plants, typically restricted to agricultural automation activities such as harvest, ...

robotic-sensing-and-stimuli-provision-for-guided-plant-growth

Research

JoVE Journal

Engineering

7.9K Views.  University of Lübeck. Distributed robot nodes provide sequences of blue light stimuli to steer the growth trajectories of climbing plants. By activating natural phototropism, the robots guide the plants through binary left-right decisions, growing them into predefined patterns that by contrast are not possible when the robots are dormant.

Video: Robotic Sensing and Stimuli Provision for Guided Plant Growth

Terahertz Microfluidic Sensing Using a Parallel-plate Waveguide Sensor

Refractive index (RI) sensing is a powerful noninvasive and label-free sensing technique for the identification, detection and monitoring of microfluidic samples with a wide range of possible sensor designs such as interferometers and resonators 1,2. Most of the existing RI sensing applications focus on biological materials in aqueous solutions in visible and IR frequencies, such as DNA hybridization and genome sequencing. At terahertz frequencies, applications include quality control, monitoring of industrial processes and sensing and detection applications involving nonpolar materials.
Several potential designs for refractive index sensors in the terahertz regime exist, including photonic crystal waveguides 3, asymmetric split-ring resonators 4, and photonic band gap structures integrated into parallel-plate waveguides 5. Many of these designs are based on optical resonators such as rings or cavities. The resonant frequencies of these structures are dependent on the refractive index of the material in or around the resonator. By monitoring the shifts in resonant frequency the refractive index of a sample can be accurately measured and this in turn can be used to identify a material, monitor contamination or dilution, etc.
The sensor design we use here is based on a simple parallel-plate waveguide 6,7. A rectangular groove machined into one face acts as a resonant cavity (Figures 1 and 2). When terahertz radiation is coupled into the waveguide and propagates in the lowest-order transverse-electric (TE1) mode, the result is a single strong resonant feature with a tunable resonant frequency that is dependent on the geometry of the groove 6,8. This groove can be filled with nonpolar liquid microfluidic samples which cause a shift in the observed resonant frequency that depends on the amount of liquid in the groove and its refractive index 9.
Our technique has an advantage over other terahertz techniques in its simplicity, both in fabrication and implementation, since the procedure can be accomplished with standard laboratory equipment without the need for a clean room or any special fabrication or experimental techniques. It can also be easily expanded to multichannel operation by the incorporation of multiple grooves 10. In this video we will describe our complete experimental procedure, from the design of the sensor to the data analysis and determination of the sample refractive index.

The procedure for implementing a refractive index sensor for terahertz frequencies based on a grooved parallel-plate waveguide geometry is described here. The method yields a measurement of the refractive index of a small volume of liquid through monitoring of the shift in the resonant frequency of the waveguide structure

Refractive index (RI) sensing is a powerful noninvasive and label-free sensing technique for the identification, detection and monitoring of microfluidic ...

Synthesis and Operation of Fluorescent-core Microcavities for Refractometric Sensing

This paper discusses fluorescent core microcavity-based sensors that can operate in a microfluidic analysis setup. These structures are based on the formation of a fluorescent quantum-dot (QD) coating on the channel surface of a conventional microcapillary. Silicon QDs are especially attractive for this application, owing in part to their negligible toxicity compared to the II-VI and II-VI compound QDs, which are legislatively controlled substances in many countries. While the ensemble emission spectrum is broad and featureless, an Si-QD film on the channel wall of a capillary features a set of sharp, narrow peaks in the fluorescence spectrum, corresponding to the electromagnetic resonances for light trapped within the film. The peak wavelength of these resonances is sensitive to the external medium, thus permitting the device to function as a refractometric sensor in which the QDs never come into physical contact with the analyte. The experimental methods associated with the fabrication of the fluorescent-core microcapillaries are discussed in detail, as well as the analysis methods. Finally, a comparison is made between these structures and the more widely investigated liquid-core optical ring resonators, in terms of microfluidic sensing capabilities.

Fluorescent-core microcavity sensors employ a high-index quantum-dot coating in the channel of silica microcapillaries. Changes in the refractive index of fluids pumped into the capillary channel cause shifts in the microcavity fluorescence spectrum that can be used to analyze the channel medium.

This paper discusses  fluorescent core microcavity-based sensors that can operate in a microfluidic  analysis setup. These structures are based on the ...

A Testing Platform for Durability Studies of Polymers and Fiber-reinforced Polymer Composites under Concurrent Hygrothermo-mechanical Stimuli

The durability of polymers and fiber-reinforced polymer composites under service condition is a critical aspect to be addressed for their robust designs and condition-based maintenance. These materials are adopted in a wide range of engineering applications, from aircraft and ship structures, to bridges, wind turbine blades, biomaterials and biomedical implants. Polymers are viscoelastic materials, and their response may be highly nonlinear and thus make it challenging to predict and monitor their in-service performance. The laboratory-scale testing platform presented herein assists the investigation of the influence of concurrent mechanical loadings and environmental conditions on these materials. The platform was designed to be low-cost and user-friendly. Its chemically resistant materials make the platform adaptable to studies of chemical degradation due to in-service exposure to fluids. An example of experiment was conducted at RT on closed-cell polyurethane foam samples loaded with a weight corresponding to ~50% of their ultimate static and dry load. Results show that the testing apparatus is appropriate for these studies. Results also highlight the larger vulnerability of the polymer under concurrent loading, based on the higher mid-point displacements and lower residual failure loads. Recommendations are made for additional improvements to the testing apparatus.

The durability of polymers and fiber-reinforced polymer composites in service is a critical aspect for their designs and condition-based maintenance. We present a novel low-cost laboratory testing platform for the investigation of the influence of concurrent mechanical and environmental loadings, and may help design more efficient yet safer composite structures.

The durability of polymers and fiber-reinforced polymer composites under service condition is a critical aspect to be addressed for their robust designs ...

Femtosecond Laser Filaments for Use in Sub-Diffraction-Limited Imaging and Remote Sensing

Probing remote matter with laser light is a ubiquitous technique used in circumstances as diverse as laser-induced breakdown spectroscopy and barcode scanners. In classical optics, the intensity that can be brought to bear on a remote target is limited by the spot size of the laser at the distance of the target. This spot size has a lower bound determined by the diffraction limit of classical optics. However, amplified femtosecond laser pulses generate intensity sufficient to modify the refractive index of the ambient air and undergo self-focusing. This self-focusing effect leads to the generation of highly intense laser filaments which maintain their intensity and small sub-millimeter diameter size at distances well beyond the classical Rayleigh length. Such intensity provides the capability of remote scanning, imaging, sensing, and spectroscopy with enhanced spatial resolution. We describe a technique for generating filaments with a femtosecond regenerative chirped-pulse amplifier, and for using the resulting filament to conduct imaging and spectroscopic measurements at remote distances of at least several meters.

High-intensity femtosecond pulses of laser light can undergo cycles of Kerr self-focusing and plasma defocusing, propagating an intense sub-millimeter-diameter beam over long distances. We describe a technique for generating and using these filaments to perform remote imaging and sensing beyond the classical diffraction limits of linear optics.

Probing remote matter with laser light is a ubiquitous technique used in circumstances as diverse as laser-induced breakdown spectroscopy and barcode ...

Design and Implementation of a Bespoke Robotic Manipulator for Extra-corporeal Ultrasound

With the potential for high precision, dexterity, and repeatability, a self-tracked robotic system can be employed to assist the acquisition of real-time ultrasound. However, limited numbers of robots designed for extra-corporeal ultrasound have been successfully translated into clinical use. In this study, we aim to build a bespoke robotic manipulator for extra-corporeal ultrasound examination, which is lightweight and has a small footprint. The robot is formed by five specially shaped links and custom-made joint mechanisms for probe manipulation, to cover the necessary range of motion with redundant degrees of freedom to ensure the patient&#39;s safety. The mechanical safety is emphasized with a clutch mechanism, to limit the force applied to patients. As a result of the design, the total weight of the manipulator is less than 2 kg and the length of the manipulator is about 25 cm. The design has been implemented, and simulation, phantom, and volunteer studies have been performed, to validate the range of motion, the ability to make fine adjustments, mechanical reliability, and the safe operation of the clutch. This paper details the design and implementation of the bespoke robotic ultrasound manipulator, with the design and assembly methods illustrated. Testing results to demonstrate the design features and clinical experience of using the system are presented. It is concluded that the current proposed robotic manipulator meets the requirements as a bespoke system for extra-corporeal ultrasound examination and has great potential to be translated into clinical use.

This paper introduces the design and implementation of a bespoke robotic manipulator for extra-corporeal ultrasound examination. The system has five degrees of freedom with lightweight joints made by 3D printing and a mechanical clutch for safety management.

With the potential for high precision, dexterity, and repeatability, a self-tracked robotic system can be employed to assist the acquisition of real-time ...

A Silicon-tipped Fiber-optic Sensing Platform with High Resolution and Fast Response

In this article, we introduce an innovative and practically promising fiber-optic sensing platform (FOSP) that we proposed and demonstrated recently. This FOSP relies on a silicon Fabry-Perot interferometer (FPI) attached to the fiber end, referred to as Si-FOSP in this work. The Si-FOSP generates an interferogram determined by the optical path length (OPL) of the silicon cavity. Measurand alters the OPL and thus shifts the interferogram. Due to the unique optical and thermal properties of the silicon material, this Si-FOSP exhibits an advantageous performance in terms of sensitivity and speed. Furthermore, the mature silicon fabrication industry endows the Si-FOSP with excellent reproducibility and low cost toward practical applications. Depending on the specific applications, either a low-finesse or high-finesse version will be utilized, and two different data demodulation methods will be adopted accordingly. Detailed protocols for fabricating both versions of the Si-FOSP will be provided. Three representative applications and their according results will be shown. The first one is a prototype underwater thermometer for profiling the ocean thermoclines, the second one is a flow meter to measure flow speed in the ocean, and the last one is a bolometer used for monitoring exhaust radiation from magnetically confined high-temperature plasma.

This work reports an innovative silicon-tipped fiber-optic sensing platform (Si-FOSP) for high-resolution and fast-response measurement of a variety of physical parameters, such as temperature, flow, and radiation. Applications of this Si-FOSP span from oceanographic research, mechanical industry, to fusion energy research.

In this article, we introduce an innovative and practically promising fiber-optic sensing platform (FOSP) that we proposed and demonstrated recently. This ...

Infrared Degenerate Four-wave Mixing with Upconversion Detection for Quantitative Gas Sensing

We present a protocol for performing gas spectroscopy using infrared degenerated four-wave mixing (IR-DFWM), for the quantitative detection of gas species in the ppm-to-single-percent range. The main purpose of the method is the spatially resolved detection of low-concentration species, which have no transitions in the visible or near-IR spectral range that could be used for detection. IR-DFWM is a nonintrusive method, which is a great advantage in combustion research, as inserting a probe into a flame can change it drastically. The IR-DFWM is combined with upconversion detection. This detection scheme uses sum-frequency generation to move the IR-DFWM signal from the mid-IR to the near-IR region, to take advantage of the superior noise characteristics of silicon-based detectors. This process also rejects most of the thermal background radiation. The focus of the protocol presented here is on the proper alignment of the IR-DFWM optics and on how to align an intracavity upconversion detection system.

Here, we present a protocol to perform sensitive, spatially resolved gas spectroscopy in the mid-infrared region, using degenerate four-wave mixing combined with upconversion detection.

We present a protocol for performing gas spectroscopy using infrared degenerated four-wave mixing (IR-DFWM), for the quantitative detection of gas species ...

Electroantennography-based Bio-hybrid Odor-detecting Drone using Silkmoth Antennae for Odor Source Localization

Small drones with chemical or biosensor devices that can detect airborne odorant molecules have attracted considerable attention owing to their applicability in environmental and security monitoring and search-and-rescue operations. Small drones with commercial metal-oxide-semiconductor (MOX) gas sensors have been developed for odor source localization; however, their real-time-odor-detection performance has proven inadequate. However, biosensing technologies based on insect olfactory systems exhibit relatively high sensitivity, selectivity, and real-time response with respect to odorant molecules compared to commercial MOX gas sensors. In such devices, excised insect antennae function as portable odorant biosensor elements and have been found to deliver excellent sensing performance. This study presents experimental protocols for odorant-molecule detection in the air using a small autonomous bio-hybrid drone based on a mountable electroantennography (EAG) device incorporating silkmoth antennae.
We developed a mountable EAG device including sensing/processing parts with a Wi-Fi module. The device was equipped with a simple sensor enclosure to enhance the sensor directivity. Thus, odor source localization was conducted using the spiral-surge algorithm, which does not assume an upwind direction. The experimental bio-hybrid odor-detecting drone identified real-time odorant-concentration differences in a pseudo-open environment (outside a wind tunnel) and localized the source. The developed drone and associated system can serve as an efficient odorant molecule-detection tool and a suitable flight platform for developing odor source localization algorithms owing to its high programmability.

This study introduces experimental protocols for a bio-hybrid odor-detecting drone based on silkmoth antennae. The operation of an experimental electroantennogram device with silkmoth antennae is presented, in addition to the structure of a bio-hybrid drone designed for odor source localization using the spiral-surge algorithm.

Small drones with chemical or biosensor devices that can detect airborne odorant molecules have attracted considerable attention owing to their ...

Mechanoluminescent Visualization of Crack Propagation for Joint Evaluation

In this study, methods for the mechanoluminescent (ML) visualization of crack propagation and mechanical behavior to evaluate adhesive joints are demonstrated and explained. The first step involved sample preparation; an air spray was used to apply ML paint to the surface of the adhesive joint specimens. The performance of the ML sensor was described to examine the measurement conditions. The results of ML sensing during a double cantilever beam (DCB) test and a lap-shear (LS) test are demonstrated as these are the most frequently and widely used methods for evaluating adhesives. Originally, it was difficult to directly quantify the crack tip and strain/stress distribution and concentration because the crack tip was too small, and the effects of the strain could not be observed. The mechanoluminescence, crack propagation, and mechanical behavior during mechanical testing can be visualized via the ML pattern during the adhesive evaluation. This allows for the recognition of the precise position of the crack tips and other mechanical behaviors related to structural failure.

In this study, a protocol is presented that describes the use of mechanoluminescent (ML) visualization for monitoring crack propagation and mechanical behavior during adhesive joint evaluation testing.

In this study, methods for the mechanoluminescent (ML) visualization of crack propagation and mechanical behavior to evaluate adhesive joints are ...

Four-Dimensional Printing of Stimuli-Responsive Hydrogel-Based Soft Robots

The present protocol describes the creation of four-dimensional (4D), time-dependent, shape-changeable, stimuli-responsive soft robots using a three-dimensional (3D) bio-printing method. Recently, 4D printing techniques have been extensively proposed as innovative new methods for developing shape-transformable soft robots. In particular, 4D time-dependent shape transformation is an essential factor in soft robotics because it allows effective functions to occur at the right time and place when triggered by external cues, such as heat, pH, and light. In line with this perspective, stimuli-responsive materials, including hydrogels, polymers, and hybrids, can be printed to realize smart shape-transformable soft robotic systems. The current protocol can be used to fabricate thermally responsive soft grippers composed of N-isopropylacrylamide (NIPAM)-based hydrogels, with overall sizes ranging from millimeters to centimeters in length. It is expected that this study will provide new directions for realizing intelligent soft robotic systems for various applications in smart manipulators (e.g., grippers, actuators, and pick-and-place machines), healthcare systems (e.g., drug capsules, biopsy tools, and microsurgeries), and electronics (e.g., wearable sensors and fluidics).

This manuscript describes a 4D printing strategy for fabricating intelligent stimuli-responsive soft robots. This approach can provide the groundwork to facilitate the realization of intelligent shape-transformable soft robotic systems, including smart manipulators, electronics, and healthcare systems.