| Περίληψη: | Industrial environments are characterized as complex, unstructured, and crowded. The local perception of the robot is limited, having a big impact on its performance (e.g., operation speed, path replanning) and safety (e.g. unexpected elements crossing through the robot’s path or found after a turning). In the implementation of hybrid production systems, workspace sharing introduces mandatory and challenging safety aspects. Any place in which autonomous robots work with humans requires constant surveillance, to guarantee that no operator nearby is put in danger by the robot. It is important that monitoring applications ensure that the position of the robot with respect to humans is always known and with high accuracy and a fast update rate. What is more, for more human safety-centered and collision preventive robot motion strategy, the forecasting of the future human movement could be utilized by and integrated with the system’s motion planner increasing the safety potential of the overall system.
In the present thesis, the integration of a workspace monitoring system is described that guarantees the safety of the operators and detects their position within a human-robot workspace, using a sensor network that consists of safety-certified devices for machinery safety and depth sensors. Finally, a prediction model is implemented based on neural networks, which can predict the future trajectory of the operator, exploiting data received from the sensor network.
|
|---|
