Automation av skogsmaskin med enbart hydrauliska sensorer
- Diarienummer
- SM07-0035
- Start- och slutdatum
- 080204-091231
- Beviljat belopp
- 651 000 kr
- Förvaltande organisation
- Umeå University
- Forskningsområde
- Informations-, kommunikations- och systemteknik
Summary
The Swedish forestry industry has a long-term goal of producing harvesting machines which use the world’s-best-practice control systems and automation engineering to increase the speed and efficiency of harvesting and reduce driver training time, with an eventual target of fully autonomous machines. Preliminary studies at the Smart Crane Lab at Umeå University have shown that automation of these heavy duty hydraulic cranes is feasible. There is, however, a barrier to the commercial application of this work: all present automation systems are totally reliant on data from optical encoders, which measure the angles and extensions of the various links of the crane, and are used in feedback control systems. However, optical encoders are too fragile to be used in real forest operations, and making them sufficiently robust is prohibitively expensive. Forestry machines must operate reliably in rough conditions. Ideally, the crane boom would be completely free of electronic devices. There is an alternative: the valves which control the hydraulic cylinders are located at the base of the crane, and may be fitted with sensing hardware without risk of damage. The proposed project is to investigate the feasibility of automating a forestry crane using only sensors on the hydraulic valves, and to develop a prototype system on a Komatsu Forest vehicle. The aim is to have a working prototype in one year, and to be able to bring a product to market soon after that.
Populärvetenskaplig beskrivning
The Swedish forestry industry has a long-term goal autonomous and semi-autonomous forestry machines. Preliminary studies at the Smart Crane Lab at Umeå University have shown that automation of these heavy duty hydraulic cranes is feasible. There is, however, a barrier to the commercial application of this work: all present automation systems are totally reliant on data from optical encoders, which measure the angles and extensions of the various links of the crane, and are used in feedback control systems. However, optical encoders are too fragile to be used in forest operations, and making them sufficiently durable is too expensive. Forestry machines must operate reliably in rough conditions. Ideally, the crane would be completely free of electronic devices. There is an alternative: the valves which control the hydraulic cylinders are located at the base of the crane, and may be fitted with sensing hardware without risk of damage. The proposed project is to investigate the feasibility of automating a forestry crane using only sensors on the hydraulic valves, and to develop a prototype system on a Komatsu Forest vehicle. The main technical challenge is to accurately recover the motion of the crane from measurements of the hydraulic system. This is an example of an observer design problem, wherein one must use knowledge of dynamical relationships to recover “hidden” variables from measurable variables. The first approaches for this sort of problem developed in the 1960s and 70s for control of aircraft and spacecraft. Since then, there has been a profusion of academic research with an array of new methods proposed, as well as many successful applications such diverse areas as biomedical engineering, process control, and stock market analysis. Collaboration between university and industry on this would be highly beneficial to both parties. For the university researcher, each new practical application brings with it new challenges, and sheds new light on the strengths and weaknesses of the observer design methods proposed in academic journals. For the industry partner, the transfer of knowledge on cutting-edge engineering techniques can offer a significant competitive advantage. Furthermore, the University has advanced rapid prototyping equipment for designing and testing new control systems, as well as the experience required to make full use of it.