UC Riverside

Development of autonomous robots that can leverage physical interactions with the environment has been increasingly attracting interest owing to the potential across applications such as precision agriculture, last-mile delivery, automated warehouses, search-and-rescue, and environmental monitoring. In addition to compliant control of rigid joints, robots with compliant or variable-stiffness bodies offer benefits such as reducing impact and adapting to different environments. This dissertation contributes to fundamental theory and practical algorithms for studying compliant aerial and legged robots capable of physical interactions with their operating environment or other robots.

The dissertation spans over modeling, control, and motion planning of aerial and legged robots, specifically focusing on stabilizing from high-speed collisions, catching flying targets, and traversing rough terrain. In the first part, we introduce resilient aerial robots equipped with compliant arms to minimize impact and detect contact. We study the dynamic modeling of the compliant vehicle when experiencing external impacts and propose a recovery method to promptly stabilize the vehicle from high-speed collisions. Furthermore, we present a collision-inclusive planning method that prioritizes contacts to facilitate navigation of aerial robots in partially-known cluttered environments. Simulated and physical experiments are conducted to demonstrate the advantages of the robot and the effectiveness of the proposed planner.The second part focuses on robot-robot interactions by presenting a solution for safely catching an aerial micro-robot in mid-air using another aerial robot equipped with a universal soft gripper. In addition to modeling and controlling the aerial vehicle, we study a planning method to avoid aerodynamic disturbances that could destabilize flying targets. Experimental results showcasing the safe capture of static and moving aerial targets are presented to demonstrate the efficacy of the approach. The third part extends the study to soft legged robots designed to traverse challenging terrains. We introduce a novel soft hexapedal robot and develop an efficient gait for overcoming rough terrain. We propose a static model for feedforward position control and present a pressure feedback controller and a closed-loop variable-height trajectory tracking method. These advancements aim to enhance the overall performance and versatility of the robot in various real-world scenarios.