CONTROL AND COORDINATION OF SUPERNUMERARY ROBOTIC LIMBS BASED ON HUMAN MOTION DETECTION AND TASK PETRI NET MODEL

This paper presents a task model and communication method used to control and coordinate a wearable robot, termed Supernumerary Robotic Limb (SRL), with a human worker during the execution of a specialized task. When controlling a collaborative system like this, we need strong communication between the robot and its wearer in order to be able to coordinate their actions. We address the communication challenges between the human worker and the SRL by monitoring the worker's actions with wearable sensors. Combining these wearable sensors together with a well defined task model allows the robot to act according to the wearer's intent. The task model is structured using Coloured Petri Nets (CPN) due to the process' deterministic and concurrent nature. We performed various tests in which the user had to execute a task while wearing the sensor suit. This data was used to establish the threshold values for our predetermined gestures and postures of interest. Detecting these postures and gestures are used to trigger task transitions in the CPN model. This allows the wearer to communicate his intentions effectively to the SRL and execute the task in a well-structured and coordinated manner with the SRL.