Aggressive driving with model predictive path integral control

Williams, Grady; Drews, Paul; Goldfain, Brian; Rehg, James M.; Theodorou, Evangelos A.

doi:10.1109/icra.2016.7487277

Cited by 302 publications

(237 citation statements)

References 18 publications

Supporting

Mentioning

233

Contrasting

Order By: Relevance

“…This platform is approximately 1 meter long, weighs over 20 kilograms, and has a top speed over 20 m/s. Previous works have demonstrated that the MPPI controller (with tuned soft cost terms) is capable of navigating this type of vehicle around a simple elliptical track [25,26], which we did our best to match in our simulation experiments. Our real-world experiments use the same type of vehicle as these prior works, but in a more challenging environment (Fig.…”

Section: /5 Scale Autonomous Racing Experimentsmentioning

confidence: 78%

“…For example, a number of sampling based methods have been derived using a bayesian approximate inference approach to stochastic optimal control [19,13], path integral control theory [21,8,5,25], and the cross-entropy method [4,24,10,11]. Despite all of the success in these areas, on-line sampling of trajectories with un-stable, non-linear dynamics in the presence of disturbances remains a key problem, and is usually addressed via ad-hoc cost function tuning.…”

Section: Related Workmentioning

confidence: 99%

“…Recently, these frameworks have been applied in MPC settings [25,6,4], where they have demonstrated the ability to control high-dimensional, non-linear systems. Since these methods do not require a gradient, they can theoretically utilize very simple encodings of tasks descriptions with sparse gradient information.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Robust Sampling Based Model Predictive Control with Sparse Objective Information

Williams¹,

Goldfain²,

Drews³

et al. 2018

Robotics: Science and Systems XIV

Self Cite

View full text Add to dashboard Cite

Abstract-We present an algorithmic framework for stochastic model predictive control that is able to optimize non-linear systems with cost functions that have sparse, discontinuous gradient information. The proposed framework combines the benefits of sampling-based model predictive control with linearization-based trajectory optimization methods. The resulting algorithm consists of a novel utilization of Tube-based model predictive control. We demonstrate robust algorithmic performance on a variety of simulated tasks, and on a real-world fast autonomous driving task.

show abstract

Section: /5 Scale Autonomous Racing Experimentsmentioning

confidence: 78%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Robust Sampling Based Model Predictive Control with Sparse Objective Information

Williams¹,

Goldfain²,

Drews³

et al. 2018

Robotics: Science and Systems XIV

Self Cite

View full text Add to dashboard Cite

show abstract

“…III-B. With the estimated states, we use Model Predictive Path Integral Control (MPPI) [27] in combination with a dynamics model to optimize a sequence of actions.Our proposed dynamic model is described in Sec. III-C.…”

Section: Methodsmentioning

confidence: 99%

Self-Supervised Learning of State Estimation for Manipulating Deformable Linear Objects

Yan

Zhu

Jin

et al. 2020

IEEE Robot. Autom. Lett.

133

103

View full text Add to dashboard Cite

We demonstrate model-based, visual robot manipulation of linear deformable objects. Our approach is based on a state-space representation of the physical system that the robot aims to control. This choice has multiple advantages, including the ease of incorporating physics priors in the dynamics model and perception model, and the ease of planning manipulation actions. In addition, physical states can naturally represent object instances of different appearances. Therefore, dynamics in the state space can be learned in one setting and directly used in other visually different settings. This is in contrast to dynamics learned in pixel space or latent space, where generalization to visual differences are not guaranteed. Challenges in taking the statespace approach are the estimation of the high-dimensional state of a deformable object from raw images, where annotations are very expensive on real data, and finding a dynamics model that is both accurate, generalizable, and efficient to compute. We are the first to demonstrate self-supervised training of rope state estimation on real images, without requiring expensive annotations. This is achieved by our novel self-supervising learning objective, which is generalizable across a wide range of visual appearances. With estimated rope states, we train a fast and differentiable neural network dynamics model that encodes the physics of mass-spring systems. Our method has a higher accuracy in predicting future states compared to models that do not involve explicit state estimation and do not use any physics prior, while only using 3% of training data. We also show that our approach achieves more efficient manipulation, both in simulation and on a real robot, when used within a model predictive controller.

show abstract

“…Both of these operations can be easily parallelized on a GPU [33]. Related work in model-based control for dynamic systems has utilized linear representations (e.g., Bayesian linear regression [35]), however, to the best of our knowledge, ours is the first work to develop a model-based controller the integrates a Koopman operator representation with sampling-based optimal control.…”

Section: A Model Representation and Data-driven Approximationsmentioning

confidence: 99%

Highly Parallelized Data-Driven MPC for Minimal Intervention Shared Control

Broad¹,

Murphey²,

Argall³

2019

Robotics: Science and Systems XV

View full text Add to dashboard Cite

We present a shared control paradigm that improves a user's ability to operate complex, dynamic systems in potentially dangerous environments without a priori knowledge of the user's objective. In this paradigm, the role of the autonomous partner is to improve the general safety of the system without constraining the user's ability to achieve unspecified behaviors. Our approach relies on a data-driven, model-based representation of the joint human-machine system to evaluate, in parallel, a significant number of potential inputs that the user may wish to provide. These samples are used to (1) predict the safety of the system over a receding horizon, and (2) minimize the influence of the autonomous partner. The resulting shared control algorithm maximizes the authority allocated to the human partner to improve their sense of agency, while improving safety. We evaluate the efficacy of our shared control algorithm with a human subjects study (n=20) conducted in two simulated environments: a balance bot and a race car. During the experiment, users are free to operate each system however they would like (i.e., there is no specified task) and are only asked to try to avoid unsafe regions of the state space. Using modern computational resources (i.e., GPUs) our approach is able to consider more than 10,000 potential trajectories at each time step in a control loop running at 100Hz for the balance bot and 60Hz for the race car. The results of the study show that our shared control paradigm improves system safety without knowledge of the user's goal, while maintaining high-levels of user satisfaction and low-levels of frustration. Our code is available online at https://github.com/asbroad/mpmi shared control.

show abstract

Aggressive driving with model predictive path integral control

Cited by 302 publications

References 18 publications

Robust Sampling Based Model Predictive Control with Sparse Objective Information

Robust Sampling Based Model Predictive Control with Sparse Objective Information

Self-Supervised Learning of State Estimation for Manipulating Deformable Linear Objects

Highly Parallelized Data-Driven MPC for Minimal Intervention Shared Control

Contact Info

Product

Resources

About