Max Robotics — Global Service Robot Directory + US Certification

About

Joseph B. Collins is a pioneering researcher in the field of neural network control for dynamic robotic locomotion. His work focuses on developing adaptive control strategies for legged robots, particularly one-legged hopping machines, using multi-layer connectionist networks. Collins’ major contributions include demonstrating that neural networks can learn to stabilize and control complex, dynamic systems without prior knowledge of the robot’s dynamics—relying solely on trial-and-error learning. His most cited paper (2003, 10 citations) presents a neural network learning strategy that maintains energy levels and minimizes losses in a hopping robot, achieving stable periodic motion. Earlier foundational works (1989–1990, collectively 17 citations) established the feasibility of using artificial neural networks for adaptive control of legged locomotion. Collins’ research has been instrumental in bridging machine learning and robotics, offering early evidence that connectionist approaches could tackle real-time control challenges in unstable, high-degree-of-freedom systems. His work remains a touchstone for researchers exploring bio-inspired control and reinforcement learning in robotics.