1、 1 An automated digging control for a wheel loader Summary An Automated Digging Control System (ADCS) for a wheel loader is developed that utilizes a behavior-based control structure combined with fuzzy logic. This controller exhibits the real-time reactive responses necessary for executing digging
2、tasks in an uncertain, unstructured and dynamic excavation environment. This paper presents field test results of a prototype ADCS that was developed and implemented on a Caterpillar 980G wheel loader. Test results show that the performance of the automated system is comparable to that of an expert
3、human operator in a wide range of excavation situations. Key Words: Fuzzy behavior control； Automated digging； Robotic excavation Introduction Automating the dig component of the excavation cycle on earth moving machines such as wheel loaders, hydraulic shovels and mass-excavators, and cable shovels
4、, has many potential benefits. Typically, when these machines are used in mining or construction applications they load large quantities of material (soil, rock, etc.) into a fleet of circulating trucks. Here, digging difficulty can vary dramatically and in these difficult digging situations effecti
5、ve loading performance is only achieved by expert operators. The dig time in these situations can double or triple which significantly reduces the output of the machine. The use of an effective automated digging control system would give every machine operator the capabilities of an expert operator,
6、 and generate the following benefits. First, consistent operation over the duration of the shift, since the control system does not get tired or lose concentration. Second, improve machine availability because the controller will always operate the machine within design limits during digging. Third,
7、 reduced wheel slippage during digging. However, to achieve these benefits and also operate effectively in the harsh excavation environment, it is important that the design of an automated system meets the following criteria. The sensors and actuators used should be limited to those currently availa
8、ble on a modern loading machine. For a wheel loader this includes electro-hydraulic actuation of 2 bucket motions, bucket position sensors and measurement of a limited number of drive train parameters. Complex sensing and actuation systems may be prone to failure in the harsh environment. Next, the
9、 system should require no input from the operator related to characterizing digging difficulty. This would require operators to make a judgement concerning digging difficulty. In general, the subsurface characteristics of the material to be loaded and its potential interactions with the bucket durin
10、g digging have the greatest effect on digging difficulty. Human operators cannot see below the surface. Thus, with no operator input the automated system must be able to adjust its digging trajectory by reacting to perceived changes in digging conditions. Automatic digging control of loading machine
11、s is particularly difficult because they operate in dynamic and unstructured environments where conditions are unknown, extremely variable and difficult to detect. On the other hand, expert human operators can achieve sophisticated control of loading machines in these difficult environments. Repeate
12、d excavation experiences help the operator to learn machine operational skills and how to adapt their operating modes to the dynamic conditions. The complexities of the interactions between the excavation machine and its environment make it impractical or infeasible to develop mathematical models ty
13、pically used in traditional control paradigms. Therefore, researchers at the University of Arizona have been developing an excavation control system that utilizes excavation knowledge gathered from skilled human operators. The Control Architecture for Robotic Excavation (CARE) is a hybrid architectu
14、re that employs a behavior-based control structure. It has reactive control at the lowest level to generate primitive bucket actions, and task planning using finite state machines (FSM) that capture excavation knowledge required for behavior arbitration. Fuzzy logic combined with behavior-based cont
15、rol provide the excavation controller with the real-time reactive response necessary for digging task execution in an uncertain and dynamic environment Several years ago, the University of Arizona researchers started a project funded by Caterpillar Inc. to use CARE as the basis to develop, implement
16、 and test an Automated Digging Control System (ADCS) on a wheel loader. The implementation platform for the prototype ADCS was a Caterpillar 980G wheel loader (see Figure 1). This wheel loader weighs 29,497 kg, is 9.5 m long, 3.75 m high and has a 4.7 mbucket. The criteria listed above were used for
17、 the designing ADCS. 3 Fig. 1. The Caterpillar 980G Wheel Loader Test Platform In this paper, we show how the CARE approach has been used to develop the prototype Automated Digging Control System on the Caterpillar 980G. The ADCS utilizes only existing production sensors and actuators and has only
18、modest computational needs. The first half of the paper details the control structure of the ADCS, while the remaining sections present data from field tests. These show that the performance of the automated system is comparable to that of an expert human operator over a wide range of excavation sit
19、es. Overview of related automated digging control work The many potential applications for automated earth moving systems has attracted a significant amount of research in this area. Typically, research has fallen into two major areas: digging process modeling and planning, and automated digging. A
20、comprehensive summary of the current research in the field is given in Singh. This section concentrates on work related to the automated digging direction. In general, the simple trajectory planning and control approach is not effective, therefore several researchers measure forces during digging which are used to adjust the digging trajectory. Bullock and Huang use these forces to initiate digging trajectory actions when
