Bumping Signal Separation and Processing in a Wheeled Mobile Robot

The bump signals should be separated and identified from the acceleration information acquired by a two dimensional accelerometer mounted on a wheeled robot to make reasonable response. The first order difference method can effectively distinguish the bump signal from the accelerations caused by a slope or rolling ground. To extract the bump information from the running acceleration signals, a detecting window of 7 points are designed, in which the signals are filtered by a seven-point median filter to remove the sharp bump signal, then the data in the window are averaged to obtain normal running acceleration, finally the bump signal can be obtained by subtract the average value with original acceleration. The separated bump signal can be used to judge the occurrence of bump. The experiment shows that the bump signal can be separated effectively, and the bump direction determined by the method is satisfactory for suitable response