A Data-Driven Approach to Prediction and Optimal Bucket-Filling Control for Autonomous Excavators

Abstract

We develop a data-driven, statistical control method for autonomous excavators. Interactions between soil and an excavator bucket are highly complex and nonlinear, making traditional physical modeling difficult to use for realtime control. Here, we propose a data-driven method, exploiting data obtained from laboratory tests. We use the data to construct a nonlinear, non-parametric statistical model for predicting the behavior of soil scooped by an excavator bucket. The prediction model is built for controlling the amount of soil collected with a bucket. An excavator collects soil by dragging the bucket along the soil surface and scooping the soil by rotating the bucket. It is important to switch from the drag phase to the scoop phase with the correct timing to ensure an appropriate amount of soil has accumulated in front of the bucket. We model the process as a heteroscedastic Gaussian process based on the observation that the variance of the collected soil mass depends on the scooping trajectory, i.e. the input, as well as the shape of the soil surface immediately prior to scooping. We develop an optimal control algorithm for switching from the drag phase to the scoop phase at an appropriate time and for generating a scoop trajectory to capture a desired amount of soil with high confidence. We implement the method on a robotic excavator. Experiments show promising results. Index Terms-Data-driven control, Gaussian process, robotics in construction, mining robotics, field robots, model learning for control, optimization and optimal control, probability and statistical methods

Excavation Reinforcement Learning Using Geometric Representation

“…Excavation Planning and Learning: The majority of prior work on excavation automation explores the excavation of homogeneous materials such as soil and sand. Here we review literature on soil and sand excavation first, which could be roughly grouped into 1) control of hydraulic earthmoving machines [3,[15][16][17]; 2) trajectory-level planning [4,18]; 3) task-level planning and excavation system engineering [2,[19][20][21]. There have have been a relatively small amount of works that focus on rigid objects excavation, partly because of the complex, stochastic nature of contact dynamics.…”

“…A S an integral part of construction, mining, and various other important engineering fields, excavator operation requires skilled workers and often needs to be operated in extreme outdoor conditions that lead to injuries [1]. There have been an increasing amount of research efforts in excavation automation [2][3][4]. However, the majority of the literature only considers the excavation of homogeneous granular materials such as soil and sand, and the excavation of irregular rigid objects such as fragmented rocks is not as explored.…”

Data-Driven Reinforcement-Learning-Based Automatic Bucket-Filling for Wheel Loaders

“…In addition, the parameters of Q-learning in source tasks are partially transferred to the Q-learning of target tasks to demonstrate the transfer learning capability of the proposed approach. Considering the nonlinearity and complexity of interactions between the bucket and pile [7], the data obtained from field tests are utilized to build a nonlinear, non-parametric statistical model for predicting the state of the loader bucket in the bucket-filling process. The prediction model is used to train the Q-learning algorithm to validate the proposed algorithm.…”

Deep Learning-Based Prediction of Throttle Value and State for Wheel Loaders

“…The data-driven approach makes it possible to deal with the complex machinery dynamics [19][20][21][22] et al used the data collected from tests to construct a nonlinear, nonparametric statistical model to predict the behavior of soil excavated by an excavator bucket. Heteroscedastic Gaussian process regression is used as the prediction framework.…”