Control of the two-wheeled inverted pendulum (TWIP) robot moving on the continuous uneven ground

Uneven terrain is common in natural environment. The TWIP robot has demonstrated its rapid movement ability in even ground in previous works. This paper explores control of the TWIP robot in the continuous uneven ground except the step case. The sinusoidal function and the piece-wise linear function type ground surface are taken as an example to explore motion characteristics of robots on complex terrain. As for the robotic system, continuous or step changes of slope are the main types of disturbance to it. To cope with these disturbance, a Sliding Mode Control (SMC) is used to design the nominal controller due to its insensitivity to uncertainty. Then, the Extended Kalman Filter (EKF) is adopted to estimate the slope of unknown terrain without any ground information from sensors but according to motion state of robot and the Equilibrium Point Estimation (EPE) is then achieved based on the obtained slope angle. In simulation, the TWIP robot moves on these two types of uneven ground surface to track the reference position or velocity trajectory. The comparative experimental results have demonstrated that SMC combined with the EPE method can enable TWIP robot to locomote on some limited uneven ground with higher stability and dynamic performance, compared to the single LQR and the single SMC method. Also, the proposed algorithm has been conduted successfully on the WIP-Bot robot to travel upward and downward slope.