Development of an Autonomous Vehicle for Weed and Crop Registration

Tom Søndergaard Pedersen, Kirsten Mølgaard Nielsen, Palle Andersen, Thomas Bak, Jens Frederik Dalsgaard Nielsen

Publikation: Working paperForskning

334 Downloads (Pure)


The extension of information technology and computers on farming tools results in new possibilities for crop/weed handling. In this paper a system using an autonomous field robot (vehicle) able to make images in the field is described. In the recent farming has come to rely on intensive use of chemicals for crop protection. A way to reduce the consumption of chemicals is to use precision techniques for placing chemicals where they have an optimal effect with minimal quantity. An important part of this is to locate the weed for automatic selective spraying. A camera placed on a sprayer may be a solution but at present the image analysis technology does not have the capability for online analysis. An alternative way is to construct a weed map prior to the spraying. In order to avoid damage to the soil a light weight vehicle carrying a camera is an obvious choice. To minimize damage to the crop the free height of the vehicle being developed in this work has been set to 0.5 m. The size and weight of the vehicle implies a power consumption of max 1.5 kW making electrical motors suitable for propulsion and steering. To make the operation affordable the vehicle should be unmanned requiring a high degree of autonomy. The vehicle is part of an autonomous information system for crop and weed registration in fields which is developed at Aalborg University and The Danish Institute of Agricultural Science. The system consists of the vehicle and a stationary base station as well as a wireless communication system between the two. The base station is the main interface to the farmer. Based on farmer information a route plan for data collection is created in the base station and send to the vehicle. The vehicle collects field data using two cameras one determinates the total weed pressure and one determinates individual species. The weed data are transmitted to the base station where weed maps are calculated. The route plan is based on a grid calculated from a priori weed spatial density knowledge. The route plan consists of a set of field coordinates; at each coordinate vehicle operations are defined. This work is funded by The Danish Ministry of Food, Agriculture and Fisheries and carried out by The Danish Institute of Agricultural Science, Research Centre Bygholm and Aalborg University in corporation with Sauer Danfoss A/S, Dronningborg A/S, Hardi International A/S and Eco-Dan. The University is responsible for the construction of the electronic part and the guidance system of the vehicle and the Agricultural Institute is responsible for the mechanical part, the image processing and the route planning. This paper focuses on the control, guidance and navigation system. A prototype platform with the dimensions of approximately 1x1x1 meters for testing the control and navigation part is described. Based on the prototype test results a new platform is under construction.
Udgiver<Forlag uden navn>
StatusUdgivet - 2002