Abstract: In order to achieve accurate mechanical work in the farmland，and improve intelligent level of agricultural machinery，an accurate aided navigation system was designed.This system was composed of two subsystems：BDS（BeiDou Navigation Satellite System）navigation subsystem and 3D vision obstacle detection subsystem.The method of BDS navigation subsystem：the RTK-BDS was used for obtaining accurate location data of agricultural machinery，the data typically included latitude and longitude，speed and heading angle.Based on farmland information，BDS navigation subsystem schedule global path of farmland，calculate drift angle，driving distance，area of cultivated of agricultural machinery during driving in the real time and draw the trail.Driver could modify direction of agricultural machinery according to planning path.The 3D vision obstacle detection subsystem：the 3D camera used for obtaining depth data.It was based on principle of infrared speckle，and not affected by natural light.Visual detection method was used to detect obstacle，calculate distance from obstacles to agricultural machinery，obstacles to center line of path，and to judge whether need to adjust direction and give driver tips.Experimental results showed that BDS navigation subsystem achieved high precision，maximum error was not more than 5 cm，and precision of obstacles detection met requirements of obstacle avoidance，its maximum error was not more than 4%.So this showed that accurate auxiliary navigation system could be used to achieve precision mechanical work.