Roboticists evaluating autonomous construction machinery have benchmarked classical path planning algorithms, revealing that planar formulations dramatically cut computational planning time without compromising path length. The simulation-based study evaluated excavator end-effector motion across 940 automated trials, establishing how distinct motion planning methods affect calculation speed, trajectory geometry, tracking accuracy, and mechanical demand during constrained manipulation tasks.
Published in the Journal of Intelligent & Robotic Systems, the research investigated four classical algorithms: A*, Rapidly-exploring Random Trees, RRT*, and Probabilistic Roadmaps. Conducted by researchers including Oleg Rogov, Eetu Miettinen, Joni Hänninen, Arto Liuha, and Grzegorz Orzechowski, the study focused on a container-entry task representative of pick-and-place operations. All experiments were carried out within the NVIDIA IsaacSim physics-based simulation platform using a detailed CAD-derived rigid-body excavator model.
The research team evaluated both full three-dimensional and planar start-to-goal formulations across parametric sweep conditions. To provide a thorough comparison of algorithmic behavior, the evaluation recorded planning time alongside trajectory geometry and tracking accuracy. The authors also tracked actuator effort metrics, incorporating torque and cumulative mechanical work estimates to serve as relative indicators of the geometric demand imposed by each planned path.
The benchmark results demonstrated distinct performance differences among the four evaluated planners. For the A* algorithm, moving from the full three-dimensional configuration to the planar formulation decreased planning time from 26.0 seconds down to 1.1 seconds. This substantial computational speedup occurred with negligible change in the overall path length, indicating that dimensional reduction offers major computational benefits for excavator operations.
Conversely, conventional Rapidly-exploring Random Trees exhibited notable performance limitations during the simulated trials. The standard RRT algorithm generated motion trajectories that were more than 60 percent longer than those calculated by A*. In addition, the cumulative actuator demand imposed by conventional RRT exceeded three times the mechanical effort required by paths planned using RRT* or Probabilistic Roadmaps.
In contrast, planar RRT* and planar Probabilistic Roadmaps demonstrated superior efficiency, achieving the lowest combined actuator effort while generating compact trajectory paths. These results show that combining planar formulations with optimal sampling or roadmap structures substantially reduces physical work estimates for excavator manipulators moving in tight operational spaces.
To verify that variations in actuator effort reflected path shape rather than numerical instability, the researchers analyzed inverse kinematics diagnostics. They monitored Jacobian condition numbers and Yoshikawa manipulability indices throughout the automated benchmark. Across all tested planners, these inverse kinematics metrics varied by less than 5 percent, confirming that the observed differences were driven by trajectory geometry rather than kinematic conditioning.
Overall, the automated simulation study establishes practical selection criteria for path planning algorithms applied to autonomous construction equipment operating in constrained workspaces. By demonstrating performance trade-offs across 940 successful trials in NVIDIA IsaacSim, the authors provide a reproducible simulation foundation designed to support future sim-to-real transfer for heavy robotic manipulators on construction machinery.
Subject of Research: Path planning algorithms for autonomous excavator manipulators in physics-based simulation
Article Title: Benchmarking Classical Path Planning Algorithms for Excavator End-Effector Motion in Physics-Based Simulation
Article References: Rogov, O., Miettinen, E., Hänninen, J., Liuha, A., & Orzechowski, G. (2026). Benchmarking Classical Path Planning Algorithms for Excavator End-Effector Motion in Physics-Based Simulation. Journal of Intelligent & Robotic Systems. https://doi.org/10.1007/s10846-026-02459-w
Image Credits: AI Generated
DOI: 10.1007/s10846-026-02459-w
Keywords: autonomous excavators, path planning, motion planning, NVIDIA IsaacSim, physics-based simulation, A* algorithm, Rapidly-exploring Random Trees, RRT*, Probabilistic Roadmaps, end-effector motion, inverse kinematics, Jacobian condition number, Yoshikawa manipulability index, actuator effort, construction robotics, sim-to-real transfer, trajectory tracking
Cite Scienmag News
Katie Riggs. (November 19, 2026). Planar Path Planning Slashes Excavator Computation Time in Physics Simulation Benchmark. Scienmag. https://scienmag.com/planar-path-planning-slashes-excavator-computation-time-in-physics-simulation-benchmark/
Katie Riggs. "Planar Path Planning Slashes Excavator Computation Time in Physics Simulation Benchmark." Scienmag, 19 November 2026, https://scienmag.com/planar-path-planning-slashes-excavator-computation-time-in-physics-simulation-benchmark/. Accessed 19 November 2026.
Katie Riggs. "Planar Path Planning Slashes Excavator Computation Time in Physics Simulation Benchmark." Scienmag. November 19, 2026. https://scienmag.com/planar-path-planning-slashes-excavator-computation-time-in-physics-simulation-benchmark/








