Automation of Basic Forestry Operations

点击翻译按钮获取中文摘要

The Canadian forest products industry can greatly benefit from recent advances in automation to alleviate issues such as productivity, labour shortage, safety, and repetitive injuries. In this proposal, we aim at demonstrating the automation of basic logging operations, by leveraging and building upon the current state-of-the-art in mobile robotics and machine learning. To this effect, we have selected key elements in forestry operation, namely harvesting and forwarding as the most susceptible to benefit rapidly and realistically from the current breakthroughs in these research areas. Our proposal aim at 1) building useful 3D maps online from onboard vehicles sensors with a LiDAR and an inertial measurement unit and 2) develop and test a number of algorithms for autonomous tree harvesting and forwarding in simulation (harvesting) and with a mock-up (forwarding). Our proposal will also make fundamental scientific contributions, notably in the area of 3D mapping, 3D perception and life-long learning. Indeed, the forest environment poses interesting challenges, with the prevalence of slopes and hills, significant occlusion by foliage, and high changes of scene appearances due to seasonal effects (foliage, snow). This proposal will contribute to the effort of FORAC 4.0 research program, a research consortium aimed at improving the competitiveness of the Canadian forest industry. The research will be conducted in direct partnership with leaders of the Canadian forest industry. Our proposal will provide a unique training experience to form the next generation of roboticists and data scientists.