This work presents a novel data-driven multi-layered planning and control framework for the safe navigation of a class of unmanned ground vehicles (UGVs) in the presence of unknown stationary obstacles and additive modeling uncertainties. The foundation of this framework is a novel robust model predictive planner, designed to generate optimal collision-free trajectories given an occupancy grid map, and a paired ancillary controller, augmented to provide robustness against model uncertainties extracted from learning data. To tackle modeling discrepancies, we identify both matched (input discrepancies) and unmatched model residuals between the true and the nominal reduced-order models using closed-loop tracking errors as training data. Utilizing conformal prediction, we extract probabilistic upper bounds for the unknown model residuals, which serve to construct a robustifying ancillary controller. Further, we also determine maximum tracking discrepancies, also known as the robust control invariance tube, under the augmented policy, formulating them as collision buffers. Employing a LiDAR-based occupancy map to characterize the environment, we construct a discrepancy-aware cost map that incorporates these collision buffers. This map is then integrated into a sampling-based model predictive path planner that generates optimal and safe trajectories that can be robustly tracked by the augmented ancillary controller in the presence of model mismatches. The effectiveness of the framework is experimentally validated for autonomous high-speed trajectory tracking in a cluttered environment with four different vehicle-terrain configurations. We also showcase the framework's versatility by reformulating it as a driver-assist program, providing collision avoidance corrections based on user joystick commands.

Motion planning for autonomous robots in human-populated environments poses numerous challenges due to uncertainties in the robot's dynamics, environment, and interaction with other agents. Sampling-based MPC approaches, such as Model Predictive Path Integral (MPPI) control, have shown promise in addressing these complex motion planning problems. However, the performance of MPPI heavily relies on the choice of the sampling distribution. Existing literature often uses the previously computed input sequence as the mean of a Gaussian distribution for sampling, leading to potential failures and local minima. In this paper, we propose novel derivations of the MPPI method to enhance its efficiency, robustness, and convergence. Our approach includes a mathematical formulation allowing for arbitrary sampling distributions, addressing numerical issues, and alleviating the problem of local minima. We present an efficient importance sampling scheme that combines classical and learning-based ancillary controllers simultaneously, resulting in more informative sampling and control fusion. We demonstrate our proposed scheme's superior efficiency and robustness through experiments by handling model uncertainties, rapid environmental changes and reducing susceptibility to local minima.

Imitation learning (IL) can train computationally-efficient sensorimotor policies from a resource-intensive Model Predictive Controller (MPC), but it often requires many samples, leading to long training times or limited robustness. To address these issues, we combine IL with a variant of robust MPC that accounts for process and sensing uncertainties, and we design a data augmentation (DA) strategy that enables efficient learning of vision-based policies. The proposed DA method, named Tube-NeRF, leverages Neural Radiance Fields (NeRFs) to generate novel synthetic images, and uses properties of the robust MPC (the tube) to select relevant views and to efficiently compute the corresponding actions. We tailor our approach to the task of localization and trajectory tracking on a multirotor, by learning a visuomotor policy that generates control actions using images from the onboard camera as only source of horizontal position. Our evaluations numerically demonstrate learning of a robust visuomotor policy with an 80-fold increase in demonstration efficiency and a 50% reduction in training time over current IL methods. Additionally, our policies successfully transfer to a real multirotor, achieving accurate localization and low tracking errors despite large disturbances, with an onboard inference time of only 1.5 ms.

Deterministic model predictive control (MPC), while powerful, is often insufficient for effectively controlling autonomous systems in the real-world. Factors such as environmental noise and model error can cause deviations from the expected nominal performance. Robust MPC algorithms aim to bridge this gap between deterministic and uncertain control. However, these methods are often excessively difficult to tune for robustness due to the nonlinear and non-intuitive effects that controller parameters have on performance. To address this challenge, a unifying perspective on differentiable optimization for control is presented, which enables derivation of a general, differentiable tube-based MPC algorithm. The proposed approach facilitates the automatic and real-time tuning of robust controllers in the presence of large uncertainties and disturbances.