Small mobile robots possess myriad advantages compared to their larger counterparts.} The compact size allows them to maneuver in confined spaces for tasks such as search and rescue and precision agriculture (e.g., early-stage disease control by monitoring leaves across crops). Further, the light weight enables them to navigate environments while exerting only moderate reaction forces, a key feature for minimally invasive operations.

As robot size decreases, maintaining mobility can become a challenge especially in complex terrains, which include flowable substrates such as leaf litter, debris, and soil. Large robots (e.g., AnyMal) experience relatively deterministic substrate interactions because the robot-to-substrate aspect ratio is so large that the substrate behaves like a continuous medium. Many of these macroscale robots have achieved remarkable success in complex terrains. Conversely, microrobots (e.g., HAMR) are designed to navigate in confined spaces. However, reduced size introduces substantial challenges in terrain interactions: most microrobots cannot achieve practical-level mobility (e.g., speed on the order of 0.1 m/s) in the field/complex environment.

In quest to develop ‘small‘ highly-capable field robots, my research will focus on mesoscale robots (approximately 0.1 kg weight, 10 cm height). As demonstrated in their biological counterparts (e.g., reptiles/centipedes), mesoscale locomotors can navigate confined spaces while maintaining mobility. However, there remains a research gap to develop highly-capable mesoscale robots. Because mesoscale locomotors and the substrate are of comparable size/weight, the environment becomes a non-equilibrium complex/dynamical system. Here, the locomotor becomes part of the environment, thereby experiencing stochastic and highly coupled substrate interactions.

To fill this gap, my current research has developed theoretical foundations, such as stochastic transportation (Chong et al., 2023, Science) and geometric mechanics (Rieser* and Chong* et al., 2024, PNAS), to model mesoscale locomotion. In particular, I identified specific forms of mechanical intelligence (MI) in reptile- and centipede-like robots to facilitate effective passive response to mesoscale challenges (Chong et al., IJRR 2021a, 2021b, 2023, Chong et al., PNAS 2022, 2023). In future work, I will investigate how MI can be incorporated with computational intelligence (CI) for embodied intelligence in mesoscale robots. Specifically, I will combine centralized/decentralized models to study CI by identifying what to sense and how to respond at mesoscale. I will use geometric mechanics to study how variations in morphology (MI) can simplify CI for mesoscale capabilities (e.g., rolling, sand-swimming, and climbing). Finally, I will develop theoretical/computational tools and study the competition/equivalency in MI and CI.