High-Speed High-Accuracy Spatial Curve Tracking Using Motion Primitives in Industrial Robots

Honglu He, Chen-Lung Lu, Yunshi Wen, Glenn Saunders, Pinghai Yang, Jeffrey Schoonover, John D. Wason, Agung Julius, John T. Wen

Introduction

In modern industrial manufacturing, the demand for high-speed automation often conflicts with the requirement for sub-millimeter precision. Tasks such as laser cutting, arc welding, and spray coating—collectively known as spatial curve tracking—require robots to follow complex geometric paths at high velocities. However, achieving this balance is notoriously difficult due to the proprietary and opaque nature of industrial robot controllers. This paper presents a systematic, three-step optimization framework designed to automate the programming of motion primitives, significantly increasing path speed while maintaining strict accuracy standards.

The Challenge of "Black Box" Controllers

Industrial robot controllers are typically "black boxes." Their internal algorithms for trajectory smoothing, blending, and joint constraint handling are proprietary, making it nearly impossible for end-users to predict exactly how a robot will behave at high speeds. Consequently, engineers often rely on a grueling manual process of trial and error, eventually settling for conservative speeds to avoid excessive vibration or path deviation. This manual tuning is a major bottleneck in deploying robots for high-precision, high-throughput applications.

Methodology

The proposed framework replaces manual tuning with an automated pipeline consisting of three primary stages: redundancy resolution, motion primitive planning, and iterative waypoint adjustment.

1. Redundancy Resolution

Many industrial tasks, such as spraying or welding, only strictly require five degrees of freedom (DoF), leaving a "free" rotation around the tool axis. The researchers utilize this kinematic redundancy to optimize the robot's pose. By employing a Differential Evolution algorithm, the framework identifies the optimal tool orientation and robot configuration that maximizes joint velocity limits and avoids kinematic singularities. This ensures the robot operates in its most "athletic" configuration, allowing for higher speeds before reaching physical limits.

2. Motion Primitive Planning

Standard CAD/CAM software often defaults to a series of linear moves (MoveL), which can lead to jerky motion and frequent decelerations. The authors implement a greedy fitting algorithm that selects the most efficient sequence of motion primitives—including linear (MoveL), circular (MoveC), and joint-space (MoveJ) moves. This approach represents the desired path with the minimum number of segments necessary to stay within a defined error threshold, resulting in smoother transitions and better utilization of the controller's internal blending logic.

3. Iterative Waypoint Adjustment

To account for the opaque behavior of the internal controller, the framework treats the robot as an optimization problem. After an initial execution (either on physical hardware or a high-fidelity simulator), the system measures the tracking error. A gradient descent-based algorithm then "nudges" the commanded waypoints in the opposite direction of the measured error. This iterative feedback loop compensates for dynamic effects and controller smoothing without requiring a detailed model of the robot's internal physics.

Experimental Evaluation

The framework was validated using two challenging benchmarks:

  • Curve 1: A multi-frequency sine wave projected onto a parabolic surface, simulating high-curvature spatial paths.
  • Curve 2: The leading edge of a turbine fan blade, representing a real-world geometry used in industrial cold spray applications.

Testing was conducted across two distinct platforms: the ABB IRB 6640 (physical testbed and RobotStudio simulator) and the FANUC M710iC-70 (RoboGuide simulator). This dual-platform approach demonstrated the framework's vendor-agnostic capabilities.

Key Results and Discussion

The experimental results demonstrated a transformative improvement over standard industry practices:

  • Significant Speed Gains: On the physical ABB robot, the optimized paths achieved a 283% speed increase for Curve 1 and a 267% increase for Curve 2 compared to the baseline.
  • Maintained Precision: Despite the increased velocity, the system maintained a maximum positional error of less than 0.5 mm and an orientation error of less than 3°.
  • Efficiency through Diversity: By incorporating circular and joint moves rather than relying solely on linear segments, the framework created trajectories that the internal controllers could process more efficiently at high speeds.

Conclusion and Future Work

This research demonstrates that by intelligently resolving redundancy and using iterative feedback to "correct" commanded paths, industrial robots can operate at significantly higher speeds than currently seen in practice. The method is particularly valuable because it does not require the manufacturer to open their proprietary controller logic.

Future research will focus on integrating high-precision external sensors (such as laser scanners) for real-time waypoint adjustment and extending the framework to multi-robot systems. These advancements will be critical for complex tasks where two robots must coordinate their motion primitives to track a moving or rotating target with high precision.

Summary updated

Abstract

Industrial robots are increasingly deployed in applications requiring an end effector tool to closely track a specified path, such as in spraying and welding. Performance and productivity present possibly conflicting objectives: tracking accuracy, path speed, and motion uniformity. Industrial robots are programmed through motion primitives consisting of waypoints connected by pre-defined motion segments, with specified parameters such as path speed and blending zone. The actual executed robot motion depends on the robot joint servo controller and joint motion constraints (e.g., velocity, acceleration limits) which are largely unknown to the users. Programming a robot to achieve the desired performance today is time-consuming and mostly manual, requiring tuning a large number of coupled parameters in the motion primitives. The performance also depends on the choice of additional param-eters: possible redundant degrees of freedom, location of the target curve, and the robot configuration. This paper presents a systematic approach to optimize robot motion parameters. The approach first selects the static parameters, then chooses the motion primitives, and finally iteratively updates the waypoints to minimize the tracking error. The ultimate performance objective is to maximize the path speed subject to the tracking accuracy and speed uniformity constraints over the entire path. We have demonstrated the effectiveness of this approach both in simulation and on physical systems for ABB and FANUC robots applied to two challenging example curves. Comparing with the baseline using the current industry practice, the optimized performance shows over 100% performance improvement.
BibTeX
@inproceedings{he2023high,
  author = {Honglu He and Chen-Lung Lu and Yunshi Wen and Glenn Saunders and Pinghai Yang and Jeffrey Schoonover and John D. Wason and Agung Julius and John T. Wen},
  title = {{High-Speed High-Accuracy Spatial Curve Tracking Using Motion Primitives in Industrial Robots}},
  booktitle = {IEEE International Conference on Robotics and Automation},
  year = {2023},
  url = {https://openreview.net/forum?id=m0DVt6GHZt},
}
Figure