Collision-Freeness and Feasibility in Non-Iterative Distributed Model Predictive Control with Prediction Mismatch

Publikationstyp
Conference Paper
Publikationsdatum
2020
Sprache
English
Author
Kloock, Christine 
Werner, Herbert 
Institut
Regelungstechnik E-14 
TORE-URI
http://hdl.handle.net/11420/8431
Citation
IEEE Conference on Decision and Control (CDC 2020)
Contribution to Conference
59th IEEE Conference on Decision and Control, CDC 2020 
Publisher DOI
10.1109/CDC42340.2020.9304138
Scopus ID
2-s2.0-85099877010
This paper revisits the theory and motivation of external soft constraints to encounter feasibility problems in non-iterative distributed model predictive control that are due to prediction mismatch. Using the idea of exact penalty theory for centralized model predictive control as a starting point, an approach is developed that ensures collision-freeness between a group of homogeneous agents in an environment with obstacles, based on recursive feasibility and a conditional hard constraint. Simulation results that illustrate the approach are presented and discussed.