Autonomous Navigation of a Tracked Mobile Robot With Novel Passive Bio-Inspired Suspension

Abstract

Most mobile robots have very simple passive suspension based on conventional mass-spring-damper methods, which bring a trade-off between loading capacity and vibration isolation, or no suspension at all with only hard contact between wheels and ground. This article presents a tracked mobile robot with a unique passive suspension, which is constructed by using a bio-inspired animal-limb-like structure, to guarantee both loading capacity and vibration isolation performance. With the novel bio-inspired suspension, the loading capacity, riding comfort and obstacle negotiation capability of the robot in various rough ground environments can be significantly enhanced. However, “soft suspension” may lead to “slippery track” due to rough terrain or irregular obstacles. To solve this problem, an advanced autonomous navigation is realized. A novel double layer nonlinear model predictive control (NMPC) architecture is adopted for optimizations of trajectory tracking. The global trajectory tracking is optimized strategically in the upper layer NMPC at a low frequency, while the local dynamics such as slippage during obstacle negotiation is timely captured in the lower layer NMPC at a high frequency. Experimental results clearly demonstrate that the superior passive suspension can efficiently absorb strong shock induced by obstacle crossing and hence guarantee smooth motion, and the double layer NMPC can effectively improve transient response and tracking accuracy ideally. This article would present a benchmark result to the design and control of mobile robots working on rough grounds.