Mission planning for mining excavators – excellence in automation

Mission planning for mining excavators – excellence in automation

first_imgCRCMining’s latest research in automation mining excavators addresses the problem of planning and executing automated excavation sequences to minimise cost, with a novel multiple-level-of-detail approach. Automation of large excavators will deliver significantly higher productivity and lower duty through consistent operation. As well as removing the operator, operational variability is reduced through the elimination of human factors. Significant improvements in productivity are also expected through the application of algorithms that optimise machine operation.An automated mining excavator requires the capability to plan its work, or missions, replacing the functions currently performed by the operator in deciding where to dig from next and when and where to move to in excavating a block of material. The mission planning problem for excavators is to find the most efficient digging sequence to remove identified blocks of material. The solution of this problem involves finding the answers to three coupled questions: ‘When to move?’, ‘Where to move?’ and ‘What to dig?’. This set of three questions is called the Tactical Movement Problem or TMP.TMP is a combinatorial optimisation problem. There are many possible digging sequences that could be used to excavate a material block; the desire is to find the best of these sequences. The best sequence is that which minimises the overall cost of excavation. This cost may be the time taken to excavate the block, or the energy needed, or a hybrid cost function combining time, energy, wear on a machine and other relevant factors.CRCMining’s research addresses the problem of planning and executing excavation sequences to minimise excavation cost. The ability to solve the problem within the time normally available between digs is of crucial importance. The investigation has shown that wavelet transformation can be used to generate coarse approximations of the terrain to be excavated, which are in turn used to produce task plans rapidly. The coarse plans are used as a guide for planning at successively finer levels of detail thus maintaining the near globally optimal solution while reducing computation time.This algorithm has been implemented on a scale robotic excavator and the approach is evaluated by comparison with a greedy algorithm with five-fold improvement in time and energy to complete a given excavation task. The multiple level of detail algorithm is shown to be robust to working in a scaled mining environment. This verification of the algorithm shows that the multiple-level-of-detail algorithm meets the requirements for a fully autonomous excavation task planner.Original contributions are made toward automated excavation through the design, implementation and verification of an optimisation based tactical level planner. The general nature of the algorithm makes it suitable for implementation in other task planning problems in which a surface environment must be manipulated by a mobile robot.This work was the topic of PhD student Peter Beasley’s thesis, supervised by CRCMining’s Automation Program Leader Professor Ross McAree, recently completed in June 2013.“This work is an excellent foundation for the next step in the automation challenge and we are hoping that the industry will continue to support our efforts to develop cost reducing technology,” said McAree. “The Centre has been fortunate to have many excellent young engineers complete PhDs with us. Peter Beasley is among the most talented and I highly commend the work he has done in the conduct of his PhD. Outstanding original research on complex and challenging problem.”last_img

Leave a Reply

Your email address will not be published. Required fields are marked *