Authors:

Kristine Yoonseo Lee

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Advancements in manufacturing industries have led to a growing demand for robots due to the need to improve productivity and operational accuracy. Currently, the role of robots in manufacturing mainly focuses on automation, being only capable of performing pre- programmed tasks in structured environments. However, real-world manufacturing sites using legacy machines are often unstructured and dynamic, increasing the need for robot autonomy which allows robots to interact with their surroundings and make decisions independently. The safety of previous autonomy frameworks is usually guaranteed through strict rule-based control, but these approaches inherently limit their adaptability. We aim to tackle this problem by developing a comprehensive framework that can handle various tasks in unstructured environments. This paper proposes a consolidated framework that integrates safe-set generation and optimal path planning. Initially, safe sets are generated based on the robot's perception of its surroundings via RGB-D camera. By leveraging Implicit Neural Representations (INR) and mapping functions, our safe sets ensure complete obstacle avoidance at a collision rate of 1%. Subsequently, an optimal trajectory to the target is generated using a modified RRT* algorithm that integrates a goal-biased random sampling and a heuristic-guided rewiring process. Comparing our proposed path planning algorithm with the conventional RRT algorithm, our path possesses 21% shorter joint- space length than the original path. In parallel, a grid-based A* algorithm is also proposed for comparison. This framework successfully performs obstacle avoidance and optimal path planning while taking real-time changes into account. BRIDGES will help roots achieve full autonomy in manufacturing sites. Keywords: Digital Twin; Robotics; Manufacturing; Automation

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Kristine Yoonseo Lee