Enabling humans and robots to collaborate frees human workers from doing heavy and repetitive tasks while allowing the execution of tasks that cannot be fully automated and necessitate human cognitive skills. Such collaboration has a high potential economic impact and is expected to play a catalytic role towards reshoring production, increasing the productivity of SMEs and meeting new market requirements and consumer needs.
The overall objective of the project is to advance robot capabilities that will allow for physical Human-Robot Collaboration (HRC) at shared workspaces without fences. As a first step towards safe and fluent collaboration, the robot must be able to perceive the pose and movements of the human collaborator. During the first year, the project experimented with existing human body recognition methods and evaluated their accuracy and efficiency in realistic environments. Moreover, it investigated how human position can be integrated in the robot motion planning and collision avoidance mechanisms to allow fluent interaction
Moving forward, the project aims at developing methods that infer short-term human intentions based on their movements and thus predicting immediate-future actions. Using human representation and intention inference, the ultimate goal is to endorse robot with decision making capabilities that serve the purposes of safe and fluent collaboration.