• HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
Tuesday, November 17, 2026
BIOENGINEER.ORG
No Result
View All Result
  • Login
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
No Result
View All Result
Bioengineer.org
No Result
View All Result
Home NEWS Science News Technology

Hierarchical Planning Cuts Welding Robot Collisions on Complex Parts

by
November 12, 2026
in Technology
Reading Time: 6 mins read
0
Hierarchical Planning Cuts Welding Robot Collisions on Complex Parts

Hierarchical Planning Cuts Welding Robot Collisions on Complex Parts

Share on FacebookShare on TwitterShare on LinkedinShare on RedditShare on Telegram

Published in Discover Mechanical Engineering, the study presents a systematic offline programming solution that automatically generates collision-free welding paths from CAD models. The approach addresses a critical bottleneck in manufacturing where manual programming remains the standard for complex, multi-planar structural parts such as construction machinery chassis and crane booms. By replacing tedious teach-in methods with an automated algorithm, the team aims to reduce the time spent on programming, which currently accounts for more than 30% of total working time in such applications.

The core innovation of the method lies in a structured welding path meta-model that classifies path nodes into primary and auxiliary categories. Primary nodes represent the weld endpoints where material is deposited, while auxiliary nodes serve as safety points for collision avoidance. This classification defines four distinct path types: weld segments, retreat segments, approach segments, and inter-seam transition segments. By assigning specific planning objectives to each type, the framework moves beyond treating the welding path as a homogeneous sequence of points. This semantic structure allows the system to apply differentiated optimization strategies, balancing local motion adaptability with global path optimality.

The data acquisition process begins with the interactive extraction of weld seam information from CAD models using the OpenCASCADE geometry kernel. Engineers visually identify target weld seams and pick the adjacent joint planes via a graphical interface. The system then automatically computes the seam geometry and reference torch pose, encapsulating these parameters into a unified data structure. This interactive approach ensures flexibility for non-standard workpieces, avoiding the limitations of pre-defined feature recognition methods. The resulting data provides a complete geometric and process characterization of the welding task, serving as the input for the subsequent planning modules.

For collision avoidance, the system employs a two-stage hierarchical strategy tailored to the specific nature of each path segment. For short approach and retreat paths, the method uses local attitude adjustment. If a collision is detected, the torch orientation is rotated around the surface normal of the interfering face in small increments until a safe clearance is achieved. This local adjustment is computationally inexpensive and preserves the ideal process orientation as much as possible. For long transitional paths between different weld seams, the system initiates a global replanning framework based on the Adaptive Batch Informed Trees (ABIT*) algorithm.

The global planning phase utilizes a novel three-stage adaptive optimization objective that dynamically adjusts priorities based on the robot’s position along the path. Near the start point, the cost function emphasizes directional constraints to guide the robot toward the correct orientation. In the intermediate region, obstacle avoidance becomes the primary objective, using a dual-zone cost function that penalizes proximity to obstacles. Near the endpoint, the focus shifts to precise convergence to the target pose. This adaptive mechanism ensures that the planner balances safety, efficiency, and accuracy across the entire trajectory, addressing the limitations of single-objective planners that often fail in complex environments.

Experimental validation was conducted in a simulation environment using the Open Motion Planning Library (OMPL) and the Fast Collision Library (FCL). The study compared the proposed three-phase method against a baseline OMPL planner that optimizes path length exclusively. In a typical box-type workpiece scenario, the proposed method reduced the average path length by 20%, from 2.45 meters to 1.95 meters. Simultaneously, the average minimum clearance between the torch and obstacles increased by 8%, from 1.47 centimeters to 1.59 centimeters. These improvements indicate that the method generates shorter, smoother paths that maintain a greater safety margin from the workpiece and fixtures.

Despite the improvements in path quality, the single-attempt success rate of the three-phase method was lower than the baseline. In the primary test scenario, the success rate was 52%, compared to 97% for the baseline method. The study attributes this trade-off to the multi-objective nature of the cost function, which prioritizes safety and smoothness over rapid convergence in cluttered spaces. However, the researchers noted that this limitation can be mitigated in practical applications through retry operations. With three retries, the effective success rate increased to 90.7%, a level considered acceptable for offline programming workflows where planning time is less critical than path reliability.

The planning time for the proposed method was slightly higher than the baseline, with a median increase of approximately 15%. In the test scenario, the median planning time was 1.77 seconds compared to 1.54 seconds for the baseline. The study argues that this computational cost is acceptable because welding paths are typically planned offline. Furthermore, the additional time required for retries to achieve a high success rate remains within a practically acceptable range. The method also demonstrated generality when tested on the BIT* planner, showing consistent though modest improvements in path length and clearance, confirming the robustness of the three-stage optimization approach.

To validate the practicality of the method, the planned paths were executed on a physical FANUC M-10iD/10L six-axis industrial robot. The robot successfully followed the generated trajectory without any collisions. The executed path closely matched the planned trajectory, with a maximum deviation of less than 2 millimeters. This deviation is within the industrial tolerance for welding applications, which is typically ±3 millimeters for box-type structures. The physical validation confirms that the simulation-based planning method translates effectively to real-world robotic operations, ensuring process feasibility and safety.

The study acknowledges several limitations, including the need for manual tuning of certain parameters, such as the middle-phase weight, which may not be optimal for all scenario complexities. The single-attempt success rate remains a challenge for real-time applications, although the offline context mitigates this issue. Future work will focus on developing adaptive parameter tuning mechanisms and integrating deep learning to optimize parameters from historical data. The method also aims to extend to more complex scenarios, such as multi-robot collaborative welding and curved welding seams. By providing a deployable offline programming solution, this research offers a significant step toward automating the welding of non-standard workpieces, reducing labor costs and improving manufacturing consistency.

A detailed examination of the algorithmic architecture shows how the three-stage optimization objective operates mathematically within the motion planner. The framework defines edge cost by multiplying geometric edge length by a factor that incorporates phase-specific penalty terms, preventing disproportionately large penalties from dominating short edge evaluations. Transition paths are segmented using distance thresholds measured from start and goal configurations, with the initial and final phase radii adapting proportionally to the overall path distance up to predefined maximum limits. When evaluating an edge connecting two states, the planner samples the midpoint to classify whether it falls within the initial, middle, or final region, thereby avoiding orientation bias caused by relying solely on boundary endpoints.

During the initial phase, a piecewise penalty assesses directional misalignment between the actual motion vector and the desired heading toward the destination, applying progressively steeper penalties as angular error crosses designated thresholds. This directional constraint gradually attenuates through a distance decay exponent as the torch progresses toward the phase radius boundary. In the middle phase, the planner prioritizes collision clearance using a dual-zone obstacle cost formulation. Inside an inner danger zone below a five-centimeter safety distance, cost escalates exponentially, while an outer warning zone extending to an influence distance of fifteen centimeters applies a decaying power function to smooth cost transitions. A mild global guidance coefficient provides subtle attraction toward the goal, ensuring the planner avoids aimless wandering without sacrificing obstacle clearance.

In the final phase, an attraction power function scaled by the final phase radius draws the tool directly toward the target pose, while obstacle penalty weighting is deliberately attenuated to allow flexible terminal repositioning. Beyond the primary test case, the researchers evaluated the three-phase method across a second transition scenario involving one hundred random simulation seeds. In that second setting, the approach reduced average path length by 10% and expanded minimum clearance by 6% relative to the baseline planner, corroborating the performance gains observed in initial trials while maintaining a planning timeout of 1.5 seconds.

Statistical dispersion across successful simulation runs highlighted notable differences in path consistency between the evaluated algorithms. For the primary transition scenario, the three-phase method achieved a median path length of 1.77 meters and an interquartile range spanning 1.47 to 2.26 meters, compared to a median of 1.91 meters and an interquartile range of 1.56 to 2.44 meters for the length-only baseline. While the baseline produced several outlier runs characterized by excessively long detours, the three-phase formulation exhibited a tighter distribution free of extreme paths. Safety distributions similarly favored the proposed approach, which registered a lower quartile minimum clearance of 1.44 centimeters versus 1.14 centimeters for the baseline, substantially diminishing near-collision trajectories near the physical torch contact tip dimensions.

Subject of Research: Hierarchical collision-free path planning for robotic welding of non-standard workpieces

Article Title: A hierarchical collision free path planning method for multi line spatial distributed welding of non standard workpieces

Article References: Hong, L., Chu, Y., & Wang, Z. (2026). A hierarchical collision free path planning method for multi line spatial distributed welding of non standard workpieces. Discover Mechanical Engineering, 5(1), Article 185. https://doi.org/10.1007/s44245-026-00382-w

Image Credits: AI Generated

DOI: 10.1007/s44245-026-00382-w

Keywords: robotic welding, path planning, collision avoidance, CAD automation, industrial robotics, manufacturing, ABIT* algorithm, offline programming, hierarchical, collision, free, path

News Source: Denise Maddox. (November 12, 2026). Hierarchical Planning Cuts Welding Robot Collisions on Complex Parts. Scienmag.

Tags: ABIT* algorithmCAD automationcollisioncollision avoidancefreeHierarchicalindustrial roboticsmanufacturingoffline programmingpathpath planningrobotic welding
Share12Tweet7Share2ShareShareShare1

Related Posts

Review Finds High Bias Across Artificial Intelligence Models for Lung Cancer Biomarkers

Review Finds High Bias Across Artificial Intelligence Models for Lung Cancer Biomarkers

November 17, 2026
Macadamia Shell Hard Carbon Anode Boosts Sodium Battery Cycling and Efficiency

Macadamia Shell Hard Carbon Anode Boosts Sodium Battery Cycling and Efficiency

November 17, 2026

Model Accurately Predicts Protein Food Ink Extrusion but Overestimates Fiber Force

November 17, 2026

Federated Framework Boosts Cross-Domain Accuracy While Cutting Uplink Data by 88 Percent

November 17, 2026

POPULAR NEWS

  • Review Links Hyperinsulinaemia to Molecular Pathways in Cholesterol Gallstone Formation

    Review Links Hyperinsulinaemia to Molecular Pathways in Cholesterol Gallstone Formation

    29 shares
    Share 12 Tweet 7
  • Self-Assessed Hirsutism Accurately Flags Endocrine and Metabolic Disruptions in Oligomenorrheic Women

    29 shares
    Share 12 Tweet 7
  • Genomic Radiosensitivity Metrics Fail to Predict Survival in Canine Sinonasal Tumors

    29 shares
    Share 12 Tweet 7
  • Global Review Highlights Escalating Burden and Care Deficits for Migraine Patients

    29 shares
    Share 12 Tweet 7

About

We bring you the latest biotechnology news from best research centers and universities around the world. Check our website.

Follow us

Recent News

Review Finds High Bias Across Artificial Intelligence Models for Lung Cancer Biomarkers

Trials Support Psilocybin for Transitional Cluster Headache and Migraine Treatment

Combined Nutrition and Psychosocial Care Aids Early Development in Severe Malnutrition

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 88 other subscribers
  • Contact Us

Bioengineer.org © Copyright 2023 All Rights Reserved.

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • Homepages
    • Home Page 1
    • Home Page 2
  • News
  • National
  • Business
  • Health
  • Lifestyle
  • Science

Bioengineer.org © Copyright 2023 All Rights Reserved.