Design and Evolution of Bio-Inspired and Multi-Modal Features for Robot Locomotion in Uneven Terrains

Robotic navigation in natural environments requires adaptability, resilience, and stability to address uncertain terrains. Animals and humans have evolved morphological features through natural selection, enabling robust and efficient locomotion, a capability that remains limited in robots. This thesis investigates bio-inspired and multi-modal approaches to improve quadrupedal robots’ ability to traverse rocky, scree, and mountainous terrains. Legged locomotion is particularly advantageous in such settings, as it allows selective foothold placement and leverages terrain features for stability and propulsion. While control strategies for quadrupedal robots are ever-strengthening, understanding foot-terrain interactions that underpin stability remains limited. Current state-of-the-art foot designs, such as flat or ball feet, rely on material compliance for stability but often slip on uneven and sloped surfaces due to point contact, leading to instability of the robot. To address this limitation, this research provides guidelines for designing an adaptive foot capable of interacting with the terrain to achieve a stable configuration. These guidelines allow for the exploration of various anatomical designs. Utilizing these guidelines, a mountain-goat-hoof-inspired foot was designed, incorporating roll and yaw rotations in the Fetlock and Pastern joints, respectively. Lower and upper stiffness bounds for these joints ensure adaptability and stability during high-force interactions with rough terrain. Experimental validation in laboratory tests demonstrated that hooved pads resist instability under higher loads. Iterative experiments in alpine-like environments refined the design into a polyurethane uni-body hoof. The design highlights the critical roles of the hard hoof-tip, hard side edge, and base compliance when navigating steep and uneven terrain, as validated by static load tests in alpine-like environments. Dynamic load tests using motion capture systems to register the interaction behavior demonstrated the hoof’s ability to reduce slip through a high-frequency stick-slip interaction phenomenon, which was inadequately present in the ball and cube control feet interaction. While ball foot compliance and cube foot edges provided limited slip resistance, their morphological evolution within the hoof proved critical highlighting its importance in robust foot-terrain interactions. Beyond morphology, animals benefit from compliant tendons in the leg that absorb impact forces during dynamic tasks. To replicate this, a modular actuator called EM-Act was developed with a series-elastic element for safe terrain interactions. Its modularity supports the creation of task-specific legged robots. However, legged locomotion is slow on simpler terrains. To address this, TraQuad, a hybrid tracked-legged robot, was developed to navigate loose, rugged terrains. TraQuad’s tracked feet provide high traction, while its legged configuration selects optimal contact points, enabling it to handle inclines and obstacles, such as 31° ramps and rocks with a 50° slope and 1.75 times the foot height. For structured environments like warehouses and city blocks, an omni wheeled-legged robot named OmniQuad was designed for agile mobility. While bio-inspired design principles significantly enhance stability and locomotion, the next step is to move beyond manually engineered solutions toward an evolutionary co-design framework. In nature, both morphology and control strategies evolve in synergy, enabling organisms to adapt optimally to their environments over generations. Similarly, in robotics, leveraging co-evolutionary approaches can lead to more adaptable and high-performing systems. This research suggests further exploration of a bi-level optimization framework, where the outer optimization focuses on evolving the robot's morphology, while the inner optimization tunes control strategies to maximize performance. By utilizing advanced simulation tools such as Isaac Sim, future studies could enable the improvement of robotic designs. This research advances robotic mobility by showcasing the advantages of bio-inspired designs and multi-modal locomotion for complex terrains. The findings demonstrate the potential of these approaches to enable stable and agile movement in challenging environments. Future advancements in evolutionary co-design hold significant promise for exploration, rescue missions, and environmental monitoring in hazardous terrains, where adaptability and resilience are critical.