CARE
Responsible:
PRIN (Relevant National Interest Projects)
DEIB Role: Partner
Start date: 2023-09-28
Length: 24 months
Project abstract
The objective of the CARE (Control of Assistive Robots in crowded Environments) project is to develop safe-by-design motion planning and control algorithms for assistive robots operating in crowded environments.
Examples of assistive robots are autonomous wheelchairs and robotic guides for blind and partially sighted people. These agents should be designed, among other things, to navigate to a chosen destination in a variety of indoor and outdoor possibly crowded environments like, e.g., offices, shopping malls, parks, walking/jogging tracks.
The challenges in controlling the motion of such assistive robots are 1) safely interacting with the user and 2) safely interacting with the surrounding crowd. For example, the robot should avoid maneuvers that can lead to a collision with the user or sudden change in direction that may undermine the user’s confidence in the robot. At the same time, the robot should not collide with the surrounding environment and, in particular, with the crowd therein. Besides such primary safety objectives, which are concerned with the user’s confidence and trust in his/her assistive robot, the project aims at developing motion planning and control algorithms which do not lead to over-cautious robot’s behaviors, which could undermine the user’s acceptance.
In order to achieve this goal, CARE proposes using predictive motion planning and control tools coupled with learning-based tools which predict the user’s and the surrounding crowd’s behavior. Such a problem formulation can lead to the persistent satisfactions of constraints that prevent the robots to unsafely interact with the user and the surrounding crowd.
The developed algorithms will be demonstrated in CARE with an autonomous wheelchair and with a robotic guide for blind and visually impaired persons called BUDD-e. On the one hand, the autonomous wheelchairs are currently available at UNIMORE and POLIMI, while the robotic guide for blind users consists of a self-driving last-mile delivery robot, called Yape, which is currently being customized at POLIMI in the project BUDD-e: in particular, to allow for the robot-user interaction, a novel system, called “smart tether”, has been realized.
Examples of assistive robots are autonomous wheelchairs and robotic guides for blind and partially sighted people. These agents should be designed, among other things, to navigate to a chosen destination in a variety of indoor and outdoor possibly crowded environments like, e.g., offices, shopping malls, parks, walking/jogging tracks.
The challenges in controlling the motion of such assistive robots are 1) safely interacting with the user and 2) safely interacting with the surrounding crowd. For example, the robot should avoid maneuvers that can lead to a collision with the user or sudden change in direction that may undermine the user’s confidence in the robot. At the same time, the robot should not collide with the surrounding environment and, in particular, with the crowd therein. Besides such primary safety objectives, which are concerned with the user’s confidence and trust in his/her assistive robot, the project aims at developing motion planning and control algorithms which do not lead to over-cautious robot’s behaviors, which could undermine the user’s acceptance.
In order to achieve this goal, CARE proposes using predictive motion planning and control tools coupled with learning-based tools which predict the user’s and the surrounding crowd’s behavior. Such a problem formulation can lead to the persistent satisfactions of constraints that prevent the robots to unsafely interact with the user and the surrounding crowd.
The developed algorithms will be demonstrated in CARE with an autonomous wheelchair and with a robotic guide for blind and visually impaired persons called BUDD-e. On the one hand, the autonomous wheelchairs are currently available at UNIMORE and POLIMI, while the robotic guide for blind users consists of a self-driving last-mile delivery robot, called Yape, which is currently being customized at POLIMI in the project BUDD-e: in particular, to allow for the robot-user interaction, a novel system, called “smart tether”, has been realized.
