Date of Award
Doctor of Philosophy (PhD)
The recent advances in the field of soft robotics have made autonomous soft robots working in unstructured dynamic environments a close reality. These soft robots can potentially collaborate with humans without causing any harm, they can handle fragile objects safely, perform delicate surgeries inside body, etc. In our research we focus on origami based compliant mechanisms, that can be used as soft robotic skeleton. Origami mechanisms are inherently compliant, lightweight, compact, and possess unique mechanical properties such as– multi-stability, nonlinear dynamics, etc. Researchers have shown that multi-stable mechanisms have applications in motion-sequencing applications. Additionally, the nonlinear dynamic properties of origami and other soft, compliant mechanisms are shown to be useful for ‘morphological computation’ in which the body of the robot itself takes part in performing complex computations required for its control.
In our research we demonstrate the motion-sequencing capability of multi-stable mechanisms through the example of bistable Kresling origami robot that is capable of peristaltic locomotion. Through careful theoretical analysis and thorough experiments, we show that we can harness multistability embedded in the origami robotic skeleton for generating actuation cycle of a peristaltic-like locomotion gait. The salient feature of this compliant robot is that we need only a single linear actuator to control the total length of the robot, and the snap-through actions generated during this motion autonomously change the individual segment lengths that lead to earthworm-like peristaltic locomotion gait. In effect, the motion-sequencing is hard-coded or embedded in the origami robot skeleton. This approach is expected to reduce the control requirement drastically as the robotic skeleton itself takes part in performing low-level control tasks.
The soft robots that work in dynamic environments should be able to sense their surrounding and adapt their behavior autonomously to perform given tasks successfully. Thus, hard-coding a certain behavior as in motion-sequencing is not a viable option anymore. This led us to explore Physical Reservoir Computing (PRC), a computational framework that uses a physical body with nonlinear properties as a ‘dynamic reservoir’ for performing complex computations. The compliant robot ‘trained’ using this framework should be able to sense its surroundings and respond to them autonomously via an extensive network of sensor-actuator network embedded in robotic skeleton. We show for the first time through extensive numerical analysis that origami mechanisms can work as physical reservoirs. We also successfully demonstrate the emulation task using a Miura-ori based reservoir. The results of this work will pave the way for intelligently designed origami-based robots with embodied intelligence. These next generation of soft robots will be able to coordinate and modulate their activities autonomously such as switching locomotion gait and resisting external disturbances while navigating through unstructured environments.
Bhovad, Priyanka B., "Developing Design and Analysis Framework for Hybrid Mechanical-Digital Control of Soft Robots: from Mechanics-Based Motion Sequencing to Physical Reservoir Computing" (2021). All Dissertations. 2913.