ASAP-MPC: An Asynchronous Update Scheme for Online Motion Planning with Nonlinear Model Predictive Control

TL;DR

As-Soon-As-Possible MPC (ASAP-MPC), an asynchronous update scheme for online motion planning with optimal control that abandons the idea of having to satisfy restrictive real-time update rates and that solves the optimal control problem to full convergence.

Abstract

This paper presents a Nonlinear Model Predictive Control (NMPC) scheme targeted at motion planning for mechatronic motion systems, such as drones and mobile platforms. NMPC-based motion planning typically requires low computation times to be able to provide control inputs at the required rate for system stability, disturbance rejection, and overall performance. Although there exist various ways in literature to reduce the solution times in NMPC, such times may not be low enough to allow real-time implementations. This paper presents ASAP-MPC, an approach to handle varying, sometimes restrictively large, solution times with an asynchronous update scheme, always allowing for full convergence and real-time execution. The NMPC algorithm is combined with a linear state feedback controller tracking the optimised trajectories for improved robustness against possible disturbances and plant-model mismatch. ASAP-MPC seamlessly merges trajectories, resulting from subsequent NMPC solutions, providing a smooth and continuous overall trajectory for the motion system. This frameworks applicability to embedded applications is shown on two different experiment setups where a state-of-the-art method fails: a quadcopter flying through a cluttered environment in hardware-in-the-loop simulation and a scale model truck-trailer manoeuvring in a structured lab environment.

Figures (9)

• Figure 1: Working principles of FUR-MPC (left), LUR-MPC (middle), and ASAP-MPC (right) and their corresponding timing diagrams (bottom row).
• Figure 2: Illustration of the trajectory jumping phenomenon for an aerial robot flying forwards in free space, occurring when naively implementing the methodology from Neunert et al. Neunert2016.
• Figure 3: ASAP-MPC: a solution to the trajectory jumping problem illustrated in Fig. \ref{['fig:traj_jumping']}.
• Figure 4: Illustration of the magnitude of (1) the feedback error compared to the actual trajectory and (2) the stitching strategy presented in Section \ref{['sec:ASAP_MPC']} for a dataset of the quadrotor experiments. The 95th and 99th percentile represented on this figure are respectively 0.64 cm and 2.47 cm (specific to that dataset).
• Figure 5: Illustration of the used simulation environment and the onboard processor used in the drone application experiments.