Comparison of UGV Position Estimation Equipped With GNSS-RTK, and GPS Using EKF

Solar energy is a growing sector in today's world, with newer technologies being introduced in the market to increase efficiency. One of these new technologies is bifacial panels that generate output from both sides of the PV (Photovoltaic) panels, which increases the production of solar energy per unit area. For maximum utilization of the bifacial panels, site assessments must be conducted to determine areas with the highest ground surface reflection. These assessments are generally carried out using a human operator for data collection. It becomes challenging, especially in harsh desert conditions due to high temperatures and challenging terrain. The data collected might also be affected due to several factors, including human errors. To overcome this problem, we built an autonomous UGV (Unmanned Ground Vehicle). The desert terrain is always challenging for the UGV due to increased wheel slippage resulting in disorientation and inaccurate movement. 

In this paper, we conducted outdoor experiments in desert terrain and collected the sensor data for the UGV position estimation. The UGV is equipped with sensors like standard GPS (Global Positioning System), GNSS-RTK (Global Navigation Satellite System – Real Time Kinematics), IMU (Inertial Measurement Unit) and optical encoder to aid in UGV position estimation. The UGV is tested on two different terrains (desert and concrete) to follow a parallel track search pattern. The parallel track search pattern is followed in our experiments as this is the general layout for PV panel mounting. These sensor data are then served as inputs for EKF (Extended Kalman Filter) with the GPS and GNSS-RTK as observers. The data collected is then post-processed using an EKF to estimate the UGV’s position.

Through this study, the EKF will be compared for standard GPS & GNSS-RTK to verify which performs better for the UGV’s position estimation. The UGV’s position is also validated using a drone camera system that uses an image processing technique that will help us validate the UGV’s position with the help of visible reference objects. We then compare the EKF position estimation and the drone camera validation for the UGV to find out if its position estimation is accurate.

The expected results should show the GNSS-RTK accuracy is more reliable in EKF for UGV position estimation compared with standard GPS but with a certain drawback. These include the high price of the GNSS-RTK, the time taken for the base station to meet absolute accuracy of less than 10cm, cloudy skies that reduces data capture speed received from surveying satellites.