Anti-Slip AI-Driven Model-Free Control with Global Exponential Stability in Skid-Steering Robots

Mehdi Heydari Shahna, Pauli Mustalahti, Jouni Mattila · 2025

Read Anti-Slip AI-Driven Model-Free Control with Global Exponential Stability in Skid-Steering Robots by Mehdi Heydari Shahna, Pauli Mustalahti, Jouni Mattila (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-08831.txt.

Introduction

Shahna, Mustalahti, and Mattila design a control system that keeps heavy skid-steer robots on course when their wheels slip. Such machines work off road, where uneven ground and severe slippage are common. Their 2025 paper treats all of the robot's motion terms as unknown to the controller, a hard starting point.

Neural networks inside the controller learn the unknown wheel behavior as the robot runs. A new adaptive rule adjusts for slippage and tunes the network weights online. The design aims to guarantee strong, fast settling of tracking errors in theory. Tests in simulation and on a real machine of nearly five tons, working on slippery ground, backed the approach.

Catalog record

CREATORMehdi Heydari Shahna, Pauli Mustalahti, Jouni Mattila
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE31 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-08831

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Anti-Slip AI-Driven Model-Free Control with Global Exponential Stability in Skid-Steering Robots Mehdi Heydari Shahna, Pauli Mustalahti, Jouni Mattila Abstract Undesired lateral and longitudinal wheel slippage can disrupt a mobile robot’s heading angle, traction, and, eventually, desired motion. This issue makes the robotization and accurate modeling of heavy-duty machinery very challenging because the application primarily involves off-road terrains, which are susceptible to uneven motion and severe slippage. As a step toward robotization in skid-steering heavy-duty robot (SSHDR), this paper aims to design an innovative robust model-free control system developed by neural networks to strongly stabilize the robot dynamics in the presence of a broad range of potential wheel slippages. Before the control design, the dynamics of the SSHDR are first investigated by mathematically incorporating slippage effects, assuming that all functional modeling terms of the system are unknown to the control system. Then, a novel tracking control framework to guarantee global exponential stability of the SSHDR is designed as follows: 1) the unknown modeling of wheel dynamics is approximated using radial basis function neural networks (RBFNNs); and 2) a new adaptive law is proposed to compensate for slippage effects and tune the weights of the RBFNNs online during execution. Simulation and experimental results verify the proposed tracking control performance of a 4,836 \bm{4,836} kg SSHDR operating on slippery terrain. I Introduction The past few years have been marked by significant global changes driven by events such as a worldwide pandemic, supply chain disruptions, resource shortages, and climate change. These factors have accelerated technological shifts, particularly through advancements in robotization and artificial intelligence (AI), which are transforming various business and industrial sectors [ 1 ] . One such sector undergoing transformation is the heavy-duty machinery industry, which now requires new insights into autonomy, reliability, and process efficiency for its future development [ 2 ] . Robotizing heavy-duty machinery requires advanced solutions that encompass both theoretical developments and application-oriented approaches, along with real-world experimental validations, to enhance automation and safety. As a result, the development of skid-steering heavy-duty robots (SSHDRs) without a separate steering mechanism has gained increasing demand in industries such as construction, agriculture, and mining. These robot platforms with differential drive systems can perform smallest radius turns by moving one set of wheels forward and the other in reverse, allowing them to pivot within their own length. This capability makes them highly maneuverable and ideal for compact, powerful, and agile operations in confined spaces, but also prone to sliding [ 3 ] – [ 4 ] . For instance, a four-wheel SSHDR prototype operating on rough terrain is shown in Fig. 1 . Fig. 1: A four-wheel SSHDR prototype operating on rough terrain. These SSHDRs commonly involve a continuous, real-time engagement of different stages within a framework [ 5 ] , as shown in Fig. 2 . Fig. 2: The whole framework of a four-wheel SSHDR. A representation of the off-road environment is mapped, often aided by sensors such as LiDAR, cameras, inertial measurement units, and GPS for autonomous navigation and localization. This representation is used to estimate the robot’s position and orientation, and the reference motion sector includes a path planner to determine an optimal, collision-free route from start to goal. The path is then refined by a path generator into a smooth, feasible trajectory, considering the robot’s kinematics and dynamics. Based on this, the robot’s required linear and angular velocities, from the immediate pose to the goal, are computed by kinematic control. The required robot motion is then transformed into individual left and right reference wheel motions using inverse kinematics. Integrating these complex and broad sectors, each requiring advanced methods to operate, with the expectation that they will seamlessly contribute to the final objective, is unrealistic in rough terrains. They must be carefully designed and separately tested to meet various standards before integration. Specifically, one common approach to evaluate the efficacy of dynamic controllers within the overall SSHDR framework in industries is through human-assisted observation and manual decision-making in the closed-loop control system, as shown in Fig. 3 . Fig. 3: Human-in-loop SSHDRs for the remote control verification. This allows a heavy robot to operate in challenging terrains with a high likelihood of slippage while enabling the human operator to assess obstacles along the path, the robot’s distance to the destination, and approximate motion commands in real-time. One instance of verifying execution of the control system in practice is shown in Fig. 4 . Fig. 4: Verification of the remote control system in practice; the video is available at: https://youtu.be/AQycbEVt_M8 . Dynamic controllers are the lowest executive system, responsible for robustly adjusting the multi-stage mechanical actuator mechanism to generate sufficient motion and ensuring stability and goal achievement. The actuation systems integrated into SSHDRs typically consist of multi-component mechanisms designed to transform energy from the power source into mechanical motion. This process involves energy excitation, high-torque motor operation, high-ratio gearing, and, ultimately, the generation of wheel movement. The inherent complexity and nonlinearities of this multi-stage actuation system pose significant challenges in accurately modeling the SSHDR’s dynamics [ 6 ] . Consequently, designing a model-based controller is significantly complicated by substantial modeling uncertainties and external disturbances. Hence, traditional model-free controllers remain widely used in the heavy-duty machinery industry due to their simplicity and lower computational demands, despite advancements in intelligent control methodologies. However, the performance of the common model-free controllers in higher-than-second-order systems is often limited, leading to oscillations, poor performance, or instability [ 7 ] . In addition, the safety and stability of SSHDRs on uneven terrain are significantly affected by wheel slippage, which arises from factors such as terrain irregularities, imbalanced load distribution, suboptimal wheel design, dynamic forces during movement, and limitations in the robot’s control mechanisms that may hinder timely adjustments [ 8 ] . Notably, despite the critical impact of wheel slippage, research on this issue across all types of heavy-duty robots has received limited attention over the past two decades [ 9 ] – [ 10 ] . Various terrains with differing slip ratios must be thoroughly analyzed to ensure the stability of the robot’s autonomous controller before integrating it with mapping systems and obstacle-free trajectory generators. Hence, [ 11 ] integrated a coordinated adaptive robust control scheme with a torque allocation technique to solve the chattering phenomenon and wheel slip compensation. However, it assumed the model of the studied skid-steering robot is accurately known. To improve the dynamic controller performance and tracking accuracy, [ 12 ] developed a traditional controller, the cascaded proportional–integral–derivative (PID) method, while compensating for the sideslip angles. However, most studies have been built on the assumption that all four wheels of the studied robot can be controlled independently, which makes it easier to address the slipping issue and compensate for it. This paper addresses the problem of robust stabilization for a four-wheel SSHDR system, in which each two-wheel side can be controlled with a single actuator. The system is characterized by a highly nonlinear multi-component actuation model and operates on severely slippery terrains. In contrast to [ 11 ] and [ 12 ] , the proposed control framework is entirely independent of the SSHDR model and interactive multi-stage actuation mechanisms. In addition, unlike [ 5 ] , the robustness of the proposed

Read the complete text