Task space adaptation via the learning of gait controllers of magnetic soft millirobots

Abstract

Untethered small-scale soft robots have promising applications in minimally invasive surgery, targeted drug delivery, and bioengineering applications as they can directly and non-invasively access confined and hard-to-reach spaces in the human body. For such potential biomedical applications, the adaptivity of the robot control is essential to ensure the continuity of the operations, as task environment conditions show dynamic variations that can alter the robot’s motion and task performance. The applicability of the conventional modeling and control methods is further limited for soft robots at the small-scale owing to their kinematics with virtually infinite degrees of freedom, inherent stochastic variability during fabrication, and changing dynamics during real-world interactions. To address the controller adaptation challenge to dynamically changing task environments, we propose using a probabilistic learning approach for a millimeter-scale magnetic walking soft robot using Bayesian optimization (BO) and Gaussian processes (GPs). Our approach provides a data-efficient learning scheme by finding the gait controller parameters while optimizing the stride length of the walking soft millirobot using a small number of physical experiments. To demonstrate the controller adaptation, we test the walking gait of the robot in task environments with different surface adhesion and roughness, and medium viscosity, which aims to represent the possible conditions for future robotic tasks inside the human body. We further utilize the transfer of the learned GP parameters among different task spaces and robots and compare their efficacy on the improvement of data-efficient controller learning.

Magnetic Miniature Actuators with Six‐Degrees‐of‐Freedom Multimodal Soft‐Bodied Locomotion

“…With six‐DOF control, the MMR could also control the angular displacements about its ‐axis and perform the two‐anchor crawling locomotion to ascend an inclined slope of ° (Figure 3D and Video S2, Supporting Information), and successfully execute sharp turns by rotating about its ‐axis at the junction of a confined “L”‐shaped path (Figure 3E and Video S2, Supporting Information). Although existing MMRs were steerable when they executed the two‐anchor crawling locomotion on obstacle‐free flat terrains, [ 10,79,80 ] they could not use this locomotion to concurrently accomplish the following tasks: 1) climb an inclined slope; 2) perform effective turns on sharp corners; and 3) tilt at an angle to better conform and thus crossing a narrow passage with strict shape constraints. The proposed MMR therefore demonstrated significantly higher dexterity than existing MMRs that could perform two‐anchor crawling locomotion.…”

“…The proposed MMR therefore demonstrated significantly higher dexterity than existing MMRs that could perform two‐anchor crawling locomotion. [ 10,49,73,79,80,82,84 ]…”

“…In theory, it is also possible to implement six‐DOF control on previous soft MMRs with harmonic magnetization profiles. [ 10,53,63,66,67,78–80 ] The key reason why such MMRs can only achieve five‐DOF motions previously is because their orientation is controlled solely by . [ 10,53,63,66,67,78–80 ] Unlike these previous works, here we use both and to control the orientation of the proposed MMR, and this enables us to generate a sixth‐DOF restoring torque for this soft actuator, effectively allowing it to realize six‐DOF motions.…”

“…[ 10,53,63,66,67,78–80 ] The key reason why such MMRs can only achieve five‐DOF motions previously is because their orientation is controlled solely by . [ 10,53,63,66,67,78–80 ] Unlike these previous works, here we use both and to control the orientation of the proposed MMR, and this enables us to generate a sixth‐DOF restoring torque for this soft actuator, effectively allowing it to realize six‐DOF motions. Although it is possible to implement six‐DOF control on previous soft MMRs with harmonic magnetization profiles, [ 10,53,63,66,67,78–80 ] their producible sixth‐DOF torque will be 1.41–63.9‐fold lower than our proposed actuator.…”

“…[ 10,53 ] To maximize the producible sixth‐DOF torque of our soft MMR, the phase shift angle of its magnetization profile, φ , has been optimized to (Figure 1A(ii) and Section S3, Supporting Information). With the selected φ , our proposed MMR can theoretically produce 1.41–63.9‐fold larger sixth‐DOF torque than existing soft MMRs that have a harmonic magnetization profile, i.e., those with φ of [ 63,66,67,78 ] and [ 10,53,79,80 ] (Section S3, Supporting Information). Because the magnitude of our MMR's achievable net magnetic moment, , is equal to all other soft MMRs with a harmonic magnetization profile (Section S3, Supporting Information), this implies that all of these MMRs will have the same actuation capabilities for their traditional five‐DOF motions.…”