This work deals with the problem of navigating a robot in a constrained human-like environment. We provide a method to generate a control strategy that enables the robot to proactively move in order to induce desired and socially acceptable cooperative behaviors in neighboring pedestrians. Contrary to other control strategies that simply aim to passively avoid neighboring pedestrians, this approach aims to simplify the navigation task of a robot by looking for cooperation with humans, especially in crowded and constrained environments. The co-navigation process between humans and a robot is formalized as a multi-objective optimization problem and a control strategy is obtained through the Model Predictive Control (MPC) approach. The Extended Headed Social Force Model with Collision Prediction (EHSFM with CP) is used to predict the human motion. Different social behaviors of humans when moving in a group are also taken into account. A switching strategy between purely reactive and pro active-cooperative planning depending on the evaluation of human intentions is also furnished. Validation of the proactive-cooperative planner enables the robot to generate more socially and understandable behaviors is done with different navigation scenarios.
Proactive-Cooperative Navigation in Human-Like Environment for Autonomous Robots
Paolo Salaris
Secondo
Writing – Review & Editing
;
2020-01-01
Abstract
This work deals with the problem of navigating a robot in a constrained human-like environment. We provide a method to generate a control strategy that enables the robot to proactively move in order to induce desired and socially acceptable cooperative behaviors in neighboring pedestrians. Contrary to other control strategies that simply aim to passively avoid neighboring pedestrians, this approach aims to simplify the navigation task of a robot by looking for cooperation with humans, especially in crowded and constrained environments. The co-navigation process between humans and a robot is formalized as a multi-objective optimization problem and a control strategy is obtained through the Model Predictive Control (MPC) approach. The Extended Headed Social Force Model with Collision Prediction (EHSFM with CP) is used to predict the human motion. Different social behaviors of humans when moving in a group are also taken into account. A switching strategy between purely reactive and pro active-cooperative planning depending on the evaluation of human intentions is also furnished. Validation of the proactive-cooperative planner enables the robot to generate more socially and understandable behaviors is done with different navigation scenarios.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
