Human-Robot Collaborative Path Planning with Predictive Obstacle Avoidance

Jointly planning paths for robots and humans in dynamic environments is considered an immensely complex task with regard to uncertainties in human motion patterns, thereby requiring real-time collision avoidance systems. Keeping in mind these challenges, there comes into existence in this publication an efficient, new model combining predictive systems for obstacle avoidance, along with hierarchical path planning to facilitate improved safety functions with optimized efficiency for collaborative robots. Utilizing the new model, there is the use of Gaussian Processes for Regression with biomechanical constraints to predict patterns in human motion with different predictive horizons in generating dynamic spatial risk maps for planning strategies for robot paths. A dual-level model, in particular, facilitates optimal trajectory planning for improved algorithms in RRT* algorithms, together with optimized Dynamic Window Approach modifications, to provide proactive collision avoidance with efficient computational processing in real-time with not more than 42.3ms planning latency in experimental analysis for UR5 in real-life collaborations for optimized safety improvements with overall success rates reaching 94.3% with optimized path-length reduction of 12.4% compared to traditional algorithms, Drawable.