Technical Note - A Study Of Autonomous Vehicle Technology Application In Mining

Recently, the North American mining industry experienced a severe recession, forcing managers to take dramatic steps to cut costs and compete in the difficult international market. Some of these steps were closing mines, reducing work forces, renegotiating wage agreements, and purchasing the most productive equipment. Managers are now looking beyond these traditional avenues and are focusing more on advanced technology. In the present environment, it is essential that mine operators obtain the maximum use of capital expended on equipment. However, mine workers do not obtain maximum efficiency or productivity from equipment because the adverse and hazardous mine environment impedes human performance. Also, the efficient use of large, complex machines calls for levels of precision that many times are beyond the capability of even highly trained miners. A possible solution to this problem is a machine that mines without operators - autonomous mining machines. However, numerous problems confront researchers who are attempting to develop such equipment. Some mining tasks are not always composed of a series of cyclic motions readily performed by factory floor robots. In addition, mining takes place in the geological environment where conditions are highly variable and unpredictable. As a result, these machines must be able to sense and adapt to variations in operating tasks and environment. Considerable autonomous vehicle research has been completed, especially for defense. However, autonomous vehicle (AV) technology has not advanced to practical application yet (King, 1988) and mining research is necessary in areas: representing mining specific knowledge; •analyzing and reasoning about sensor data in the mining environment; and •discovering completely new mining methods or new approaches to existing methods that become apparent when we remove the constraints imposed by the necessity of human operators. A specific machine, the LHD, can be used to show the problems for researchers who are attempting to develop autonomous machines for mining. This author chose the LHD because it can borrow concepts developed for autonomous navigation by military programs. But considerable mining research is also required. Furthermore, studies done at the Henderson mine, in Colorado, show autonomous LHD 's promise cost and safety benefits (King, 1988). At Henderson, LHDs load ore from draw points, tram to an ore pass, dump and return, or switch to another draw point. An autonomous LHD must sense vehicle position along the route, relate sensor data to stored-map information, to determine location, follow drift center lines, plan paths between dump position and initiate appropriate control commands, sense vehicle operating status and vehicle health, key on features or targets for special tasks like high speed turns, perform end of travel tasks (loading and dumping), and detect and avoid obstacles. These goals are similar to those for shuttle cars, trucks, and front-end loaders. Therefore, much of the technology is transferable. Each Henderson LHD extracts six to 40 dippers from each of a series of draw points. The LHD transports the ore to an ore pass within 55 m (180 ft) of the draw point, making the longest round trip 110 m (360 ft). The LHDs have very fast hydraulic dumping and loading systems that reduce the round trip cycle to less than one minute. Even though the LHD is capable of 500 trips per shift, the average production is 300 dippers. Man trip and lunch reduce available operating time to 6.5 hours per shift. Mucking difficulties (setting large boulders aside), operator breaks for activities like talking shop, and cleaning and smoothing roads further reduce operating time. Supervised autonomy can reduce the number of operating units by increasing operating time per shift since computer controlled machines can operate during lunch and between shifts and reduce operator errors. If dippers per shift increase from 300 to 350 (long-range goals are 500 dippers per shift), constant production requires only 10 operating LHDs and two spares. Manpower requirements drop from 24 to four by controlling five machines from one workstation. A review of technology available from the Autonomous Land Vehicle, the Advanced Ground Vehicle Technology, the Ground Surveillance Robot, and other programs, show the following differences between others work and mining industry needs: •If we focus initially on mobile haulage vehicles, we can navigate from a map. We do not need to explore. •We can modify the environment to reduce the navigation requirements. •We have a harsher environment than any of the research programs have encountered. •Our equipment must operate faster and more precisely than present AVs. •Our equipment must operate reliably over long periods of time. •We must have better onboard machine health monitoring and diagnostics. •The AV programs do not address geosensing. •The major AV programs are not cost-effective. For example, we cannot afford the computer power for robust image processing, yet. To computer control an LHD, we must replace the guidance and monitoring skills of experienced operators with sensors, computing hardware, interfaces, and several software mod¬ules. Experienced operators avoid collisions and load efficiently in piles that may contain oversize muck. Collision avoidance without an operator requires sensing all obstacles in the draw