Source-linked AI summary
ClimbLab: MATLAB Simulation Platform for Legged Climbing Robotics
Kentaro Uno, Warley F. R. Ribeiro, Yusuke Koizumi, Keigo Haji, Koki Kurihara, William Jones, Kazuya Yoshida
TL;DR
Climbing robots are difficult to test safely in harsh environments, motivating simulation for design and validation. ClimbLab is an open-source MATLAB platform for designing and simulating legged robots walking and climbing. Its examples compare cliff-holding capability and quadruped gaits, with the diagonal gait showing better minimum stability values and lower RMS torque than the normal gait.
Problem
Experiments with climbing robots are difficult on harsh terrain and can require large-scale safety equipment, while simulation supports design and high-risk validation.
Source: §1 · pp. 1–2Method
ClimbLab is an open-source standalone MATLAB simulator for legged robots that combines robot design, gait simulation, and equilibrium analysis using SpaceDyn for rigid-body dynamics.
Source: §1.2 · p. 2, §2 · pp. 2, 5Results
The diagonal gait had better minimum TSM and GIA margins than the normal gait and used 13.5% less RMS torque; the robot held a cliff inclined up to 100° but failed at 110°.
Source: §3.1 · p. 6, §3.2 · p. 7Takeaways & Limitations
The simulations support systematic gait-performance evaluation and estimation of static performance and the minimum gripping force required for a specified environment.
Source: §3.1 · p. 6, §3.2 · p. 7
Abstract
from arXiv · showhide
This paper presents an open-sourced MATLAB simulation and analysis platform dedicated to legged climbing robots. This simulator enables the design of any limbed robotic system as an articulated multi-body with a floating base and simulates it walking and climbing in an arbitrary environment. The main variable environmental parameters are inclination, gravity, and ground stiffness, and any point cloud can be installed as the terrain map. Furthermore, the simulator employs a rigid body dynamics engine. This paper first describes the simulator structure, and the computational flow and next presents the representative simulation examples where quadrupedal robots assumed gripping on the wall or climbing on the steep slope.
1 Introduction
Climbing robots can traverse steep and complex terrain, but safe, accessible experiments are difficult, making simulation important for design and validation. ClimbLab addresses a gap in platforms dedicated to legged climbing robots.
1 Introduction
- Climbing robots use foot-mounted gripping mechanisms to traverse steep slopes, overhanging walls, and ceilings, including glass, rocky, and metal surfaces.Examples include the Ninja series, Lemur series, and Magneto.Source: §1 · p. 1
- Experiments with climbing robots are difficult to conduct on harsh terrain and may require large equipment to protect robots and operators.Simulation can support robot design choices and assess climbing performance in severe or high-risk situations before deployment.Source: §1 · pp. 1–2
1.1 Related work
- Existing general-purpose MATLAB robot toolboxes support many robot types and libraries, which can make the full systems challenging to understand.MATLAB is described as offering intuitive coding for quickly expressing ideas, while other cited simulators are mainly written and supported in C/C++ and Python.Source: §1.1 · p. 2
1.2 Contribution
- ClimbLab is an open-source, standalone platform dedicated to legged robots walking and climbing, and it depends on SpaceDyn for rigid-body dynamics.SpaceDyn is installed automatically; the authors report that dedicated platforms for legged climbing robots had not yet been established to their knowledge.Source: §1.2 · p. 2
- This initial proceedings report describes ClimbLab’s simulation flow and presents representative application examples.The paper first summarizes simulator structure and computation flow, then displays samples to illustrate usage.Source: §1.2 · p. 2
2 ClimbLab
ClimbLab combines robot design, gait simulation, and equilibrium analysis in a standalone simulator. Its modular workflow covers environment sensing and planning, physics-based dynamics, visualization, and data export.
2 ClimbLab
- ClimbLab integrates robot design, gait simulation under environmental constraints, and static and dynamic equilibrium analysis in a standalone platform.The simulator structure and computational flow are organized around a main script and role-specific function files.Source: §2 · pp. 2–3
- Initialization defines the robot and terrain, sets gravity, inclination, and timing parameters, and creates files for later analysis.Implemented options include insect- and mammalian-type quadrupeds, flat or rough terrain, and rough terrain based on sensed climbing-hold data.Source: §2 · p. 3
- Environment sensing gathers observable surface information and identifies graspable-point candidates before gait and local step planning.The sensing functions can account for camera field of view and surface and gripper geometry.Source: §2 · p. 4
- Gait planning selects reachable grasp points, the next limb and foothold, and a base COM position, then checks stance feasibility.Spline trajectories can connect initial and final Cartesian positions; analytical inverse kinematics computes the required joint angles.Source: §2 · p. 4
- Equilibrium evaluation considers pose, gravity, acceleration, and gripping force, and supports TSM and GIA margins alongside classical support-polygon analysis.TSM and GIA are recommended for gripper-equipped robots because they include maximum holding force in the tumbling moment.Source: §2 · p. 4
- A PD torque controller targets desired joint positions, while a compliant contact model computes ground reaction forces in all Cartesian directions.Gripper contact can persist above the surface while grasp conditions hold, with detachment based on equilibrium criteria or force and holding-force limits.Source: §2 · p. 5
- SpaceDyn numerically integrates the floating-base rigid-body equation of motion with Runge–Kutta and also computes forward kinematics.The equation includes robot inertia, Coriolis and centrifugal effects, gravity, joint inputs, Jacobians, and ground reaction forces.Source: §2 · p. 5
- MATLAB plots display the robot and surface during simulation, and selected variables can be saved for later time-history analysis.Additional animation information can include support polygons, graspable points, reachable areas, and force or acceleration vectors.Source: Figure 1 · p. 3, §2 · p. 5
3 Application Examples
The examples use simulations to assess cliff-holding limits, compare climbing gaits, and test goal-directed foothold selection on discrete landable points.
3.1 Simulation (1) – A Robot Holding onto a Cliff
- With an assumed 13.0 N gripper, HubRobo held an overhanging uneven slope at 100° but failed at 110°, illustrating static performance and grip-force estimation.Changing gripping force rather than terrain settings can estimate the minimum force needed for a specific environment.Source: §3.1 · p. 6, Figure 3 · p. 7
3.2 Simulation (2) – Gait Analysis for Quadrupedal Robotic Climbing
- Normal gait keeps three feet supporting the body, whereas diagonal gait shifts the base to keep COM projection at the leg-diagonal intersection on steep terrain.The diagonal gait keeps the projection on the support triangle’s edge; gripping force expands the support polygon’s stability boundary.Source: §3.2 · p. 6
- 13.5% less RMS torque was used by diagonal gait; its minimum TSM and GIA margins were better, though the two gaits’ average margins were nearly unchanged.Diagonal gait also exhibited more constant stability during the simulation.Source: §3.2 · p. 7, Table 1 · p. 8
3.3 Simulation (3) – Sensing and Optimal Footholds Selection
- For random discrete landable points, kinematic-only simulation showed optimal local foothold selection could reach the goal; whole-body planning still needs stability, energy, and manipulability.The simulation omitted dynamics to reduce runtime.Source: §3.3 · p. 8
4 Conclusion
ClimbLab is a modular MATLAB framework for simulating legged robots walking and climbing, with code and documentation provided for adaptation.
4 Conclusion
- ClimbLab is a MATLAB framework for simulating legged robotic systems walking and climbing in various environments.Source: §4 · p. 8
- Systematically modularized MATLAB functions support robot design, walking and climbing operation, and analysis of kinematic and dynamic behavior.Source: §4 · p. 9
- Reference simulation code is available on GitHub for modification into custom simulations, alongside documentation and additional examples.Source: §4 · p. 9