UC San Diego

As mobile robots become increasingly prevalent in society, there is a need to design robots to be robust to deployment in natural terrain environments. Natural environments feature diverse sets of challenges that are terrain specific, for example the strategies and robot morphologies needed for movement over sand dunes differs greatly from those needed for movement through dense forest underbrush, or through caves systems. As we seek to extend the range of robots from the human engineered spaces that they mostly inhabit today, new robot designs and movement strategies are needed in order to ensure robust and efficient locomotion. To date, much work has been done on complex sensing and control strategies to compensate for irregular environmental features, but these control strategies often involve expensive components and are computationally expensive. Alternatively, we can use morphology solutions to offload computation to the body of the robot, reducing sensor needs and freeing some computational power to be used for other tasks. To achieve this, we can harness new actuation strategies, unconventional materials, and novel robot appendage designs to modulate robot-environment interactions using only basic control loops. This dissertation describes the design of robot-environment interfaces that to improve physical interactions between a robot and it’s environment in order to enable locomotion through that environment. First, I use granular jamming, a mechanism to achieve variable stiffness, and create shear-strengthened jamming robot appendages that can conform over irregular terrain. I demonstrate that, when internally reinforced with abrasive fibers, these granular jamming feet allow a hexapod robot to walk faster and with more traction over rough terrain. Next, I present a robot featuring directionally compliant telescoping appendages that allow it to passively compress to enter both vertically and horizontally confined spaces without sacrificing performance over open terrain features such as steps and rock gardens. Finally, I present a method for directional drag reduction using porous diffusive sheets and granular fluidization for improving the digging capabilities of mobile robots. Overall, these results demonstrate methods for designing robot morphologies to elicit specific interactions when moving through a target environment.