Research

Predictive Control of Soft-Growing Robots

Mimicking the biological growth exhibited by plants, soft-growing robots have shown superb performance in exploring tight and distant environments due to their flexibility and lengths that can extend up to several tens of meters. However, controlling the position of the growing robot’s tip could be challenging due to the lack of an adequate method for precisely measuring the robot’s Cartesian position in working environments. Meanwhile, owing to the irreversible addition of materials in growing robots, classical control techniques would fail to obtain feasible control actions considering the process constraints. This research proposes two optimization-based approaches, namely, a Moving Horizon Estimation (MHE) combined with a Nonlinear Model Predictive Control (NMPC) scheme to achieve the robot’s performance. The MHE estimates the robot’s entire state, including its unknown Cartesian position, based on the available robot’s configuration measurements. The proposed NMPC ensures the robot’s performance in point stabilization, trajectory tracking, and obstacle avoidance while considering the process constraints by relying on the estimated state. The nonlinear kinematic model of a vine-like robot, one of the newly introduced plant-inspired growing robots, is incorporated in numerical simulations, and satisfying results are obtained in terms of significantly reduced computation times and tracking errors.

Force and Shape Sensing of Multi-Segements Soft Continuum Robots