eduzhai > Applied Sciences > Engineering >

Risk-Averse MPC via Visual-Inertial Input and Recurrent Networks for Online Collision Avoidance

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: In this paper, we propose an online path planning architecture that extendsthe model predictive control (MPC) formulation to consider future locationuncertainties for safer navigation through cluttered environments. Ouralgorithm combines an object detection pipeline with a recurrent neural network(RNN) which infers the covariance of state estimates through each step of ourMPC s finite time horizon. The RNN model is trained on a dataset that comprisesof robot and landmark poses generated from camera images and inertialmeasurement unit (IMU) readings via a state-of-the-art visual-inertial odometryframework. To detect and extract object locations for avoidance, we use acustom-trained convolutional neural network model in conjunction with a featureextractor to retrieve 3D centroid and radii boundaries of nearby obstacles. Therobustness of our methods is validated on complex quadruped robot dynamics andcan be generally applied to most robotic platforms, demonstrating autonomousbehaviors that can plan fast and collision-free paths towards a goal point.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×