Neural Network Controller for Autonomous Pile Loading Revised

We have recently proposed two pile loading controllers that learn from human demonstrations: a neural network (NNet) [1] and a random forest (RF) controller [2]. In the field experiments the RF controller obtained clearly better success rates. In this work, the previous findings are drastically revised by experimenting summer time trained controllers in winter conditions. The winter experiments revealed a need for additional sensors, more training data, and a controller that can take advantage of these. Therefore, we propose a revised neural controller (NNetV2) which has a more expressive structure and uses a neural attention mechanism to focus on important parts of the sensor and control signals. Using the same data and sensors to train and test the three controllers, NNetV2 achieves better robustness against drastically changing conditions and superior success rate. To the best of our knowledge, this is the first work testing a learning-based controller for a heavy-duty machine in drastically varying outdoor conditions and delivering high success rate in winter, being trained in summer.

I. INTRODUCTION

Pile loading is one of the most challenging tasks in earth moving automation for Heavy-duty mobile (HDM) machines. This is partly caused by the difficulty of modelling the interaction between the tool and the material [3] and partly because of high variation in work sites and weather conditions throughout the year (Fig. 1). Weather conditions affect the material properties, the hydraulics properties of the machine, and the ground surface properties. The state-of-the-art works on pile loading or excavation automation are either model-based or use heuristics [4], [5], and experimented only in simulators or with toy setups. Therefore it is unclear how well these methods perform in real work sites.