Iterative-learning error compensation for autonomous parking of mobile manipulator in harsh industrial environment

Highlights

• High-precision and low-cost parking error measurement method.

• Iterative-learning error compensation scheme for the parking error.

• Efficacious estimation of environment fluctuations using fuzzy logic rules.

• The feasibility is verified in real manufacturing factory.

Abstract

In wide-area and multi-sites manufacturing scenarios, the mobile manipulator suffers from inadequate autonomous parking performance due to the harsh industrial environment. Instead of struggling to model various errors or calibrate multiple sensors, this paper resolves the above challenge by proposing an iterative-learning error compensation scheme that consists of offline pre-regulation and online compensation, which can improve the compensation efficiency and accommodate the error fluctuations caused by environmental fluctuations. Integrating an improved Monte-Carlo localization and eye-in-hand vision technique, an effective measurement system is firstly developed to accurately obtain the parking data without requiring superfluous facilities or cumbersome measurement. Then, after removing the data outliers utilizing the Grubbs test, offline pre-regulation is achieved to give a suitable initial value and increase the compensation convergence. To reduce the time-varying systematic errors and parking error fluctuations, online compensation is presented by offering an efficacious estimation of environmental fluctuations using fuzzy logic rules and providing an adaptive iterative-learning law. Finally, the feasibility and effectiveness of the presented compensation method are validated by extensive experiments implemented on a self-developed mobile manipulator.

Introduction

Beneficial from the locomotive flexibility and operability, mobile manipulators are generally accepted in wide-area and multi-sites manufacturing scenarios to perform transportation and manipulation [1], [2], [3], [4]. This brings notable potentials for retrofitting and renovating of the typical manufacturing workshops shown in Fig. 1. For practical manufacturing, a mobile manipulator has to navigate among multiple workbenches under specific task-scheduled requirements [5], [6]. It implies that the mobile manipulator needs to achieve autonomous parking to reach each desired site reliably and repeatedly [7], [8], [9]. However, the parking precision will be affected by the inevitable systematic errors due to the mechanical clearance, sensor errors and harsh environment, such as unsatisfactory ground conditions and dynamic obstacles [10], [11]. Consequently, to guarantee successful manufacturing with high accuracy and efficiency, it is crucial to explore an applicable method to eliminate the autonomous parking errors under fluctuating environment as the mobile manipulators go further into various applications.

For the typical manufacturing scenario demonstrated in Fig. 1, the mobile manipulators can be applied for transferring the workpiece between different sites on the pitted and greasy ground under dynamic obstacles. To make the whole workshop to be automated, we have developed a series of commercial mobile manipulators shown as Fig. 2. It consists of a four-wheel-steering and four-wheel-driving (4WS4WD) mobile platform, a manipulator, a vision system, a storage rack, eight motor encoders and related sensing devices, including, lidars, ultrasonic transducers and several anti-collision strips. As a typical modular robotic system [12,13], this robot can flexibly configure the onboard operating platform (e.g., manipulator, roller or belt conveyer and inspection equipment, etc.) as required. Besides, it has some remarkable characteristics, such as automatic charging, multiple collision avoidance, trackless autonomous navigation and vision-based operating the workpiece. With four actively steerable wheels actuated by hub motors and gearboxes, this mobile manipulator has remarkable maneuverability to perform autonomous parking and manipulation among multi-sites. Based on the above advantages, it is applicable for loading, unloading and transferring the workpieces in manufacturing workshops, which greatly improves the manufacturing flexibility and production efficiency.

Comparing to conventional differential mobile platforms running in a favorable environment [14], the difficulties in compensating the systematic parking error of the proposed system lie in: 1) The structure of the 4WS4WD mobile manipulator is more complicated and interconnected, and the coupling features between the moving platform and onboard manipulator make the calculation of the mechanical clearance and precise control hard to achieve; 2) It is cumbersome to calibrate the applied sensors accurately and uniformly; 3) In a harsh industrial environment, the error model is difficult to be built or identified accurately due to the unmodeled dynamics, uneven ground and system uncertainties.

Bearing the above-mentioned challenges in mind, an iterative-learning error compensation method is proposed in this paper to achieve precise autonomous parking of a mobile manipulator. This method simplifies the compensation process and rejects the environmental fluctuations. Without cumbersome sensor calibration and time-consuming modeling, it has three distinguish improvements: 1) With improved Monte-Carlo localization and vision-based technique, a high- precision and low-cost measurement method is designed to obtain the parking errors, which is applicable for wide-area and multi-sites industrial scenarios; 2) Using a small amount of testing data with removed outliers, the offline pre- regulation provides an optimized initial value to enhance the compensation convergence and efficiency; 3) The iterative-learning law constructed by fuzzy logic rules can be adjusted adaptively to avoid oscillation caused by error fluctuation changes and mitigate the environmental influence on the parking accuracy. Implemented in various scenarios, the practicability of our approach has been verified by autonomous parking assessments with an industrial mobile manipulator. It is concluded that the proposed method can effectively reduce the parking errors and accommodate the dynamic environment without violent oscillation.

The remainder of this paper is structured as follows. In Section 2, we provide related works. The error measurement method is given in Section 3, followed by the iterative-learning error compensation method in Section 4. Section 5 and Section 6 offer experimental validations and conclusions, respectively.