Home
Uni-Logo
 

TrimBot: a gardening robot for rose, hedge and topiary trimming

EU Grant 688007 TrimBot2020 Horizon2020

January 2016 - December 2019

Project members

Funded members:
Prof. Dr. Thomas Brox
Nikolaus Mayer

Huizhong Zhou
Eddy Ilg
Co-workers:
Tonmoy Saikia

The TrimBot2020 project will research the robotics and vision technologies to prototype the first outdoor garden trimming robot. The robot will navigate over varying terrain, approach rose bushes, hedges and boxwood topiary, to trim them to an ideal shape. The robot will be based on a modified Bosch Indego robot lawnmower, which will navigate using a user-defined garden map and 3D scene analysis, and then visually servo a novel electric plant cutter. Achieving this requires a combination of robotics and 3D computer vision research and innovation activities. Original developments are required for 3D sensing of semi-regular surfaces with physical texture (overgrown plant surfaces), coping with outdoor lighting variations, self-localising and navigating over real terrain and around obstacles, visual servoing to align the vehicle with potentially moving target plants, visual servoing to align leaf and branch cutters to a compliant surface, and innovative engineering to deliver all this on a small battery-powered consumer-grade vehicle.

The team in Freiburg is responsible particularly for accurate estimation of depth maps, optical flow, and scene flow from stereo image sequences. These tasks are approached using end-to-end training of deep networks on rendered datasets specifically designed for the challenges in the garden scenario. Moreover, the team will work on detailed modelling of the involved plants in order to allow for the planning and control of the trimming tool.

More details on the overall project can be found on the consortium page.

Publications related to this project

European Conference on Computer Vision (ECCV), 2018

Alexey Dosovitskiy, Philipp Fischer, Eddy Ilg, P. Häusser, C. Hazırbaş, V. Golkov, P. Smagt, D. Cremers, Thomas Brox
IEEE International Conference on Computer Vision (ICCV), 2015