eduzhai > Applied Sciences > Engineering >

Trajectory Generation by Chance Constrained Nonlinear MPC with Probabilistic Prediction

  • Save

... pages left unread,continue reading

Document pages: 13 pages

Abstract: Continued great efforts have been dedicated towards high-quality trajectorygeneration based on optimization methods, however, most of them do not suitablyand effectively consider the situation with moving obstacles; and moreparticularly, the future position of these moving obstacles in the presence ofuncertainty within some possible prescribed prediction horizon. To cater tothis rather major shortcoming, this work shows how a variational BayesianGaussian mixture model (vBGMM) framework can be employed to predict the futuretrajectory of moving obstacles; and then with this methodology, a trajectorygeneration framework is proposed which will efficiently and effectively addresstrajectory generation in the presence of moving obstacles, and alsoincorporating presence of uncertainty within a prediction horizon. In thiswork, the full predictive conditional probability density function (PDF) withmean and covariance is obtained, and thus a future trajectory with uncertaintyis formulated as a collision region represented by a confidence ellipsoid. Toavoid the collision region, chance constraints are imposed to restrict thecollision probability, and subsequently a nonlinear MPC problem is constructedwith these chance constraints. It is shown that the proposed approach is ableto predict the future position of the moving obstacles effectively; and thusbased on the environmental information of the probabilistic prediction, it isalso shown that the timing of collision avoidance can be earlier than themethod without prediction. The tracking error and distance to obstacles of thetrajectory with prediction are smaller compared with the method withoutprediction.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...