Method for analyzing error and existence probability of multi-sensor fusion of obstacle detection

1. A method for analyzing an error and an existence probability of a multi-sensor fusion, which is for detecting an obstacle on a travel direction of a vehicle, the method comprising:

providing a plurality of sensors to detect the obstacle to generate a plurality of observing positions of the obstacle and a plurality of observing velocities of the obstacle;

providing a processor to generate a plurality of predicting positions of the obstacle and a plurality of predicting velocities of the obstacle according to the observing positions of the obstacle and the observing velocities of the obstacle, respectively;

providing a plurality of predetermined error-average distributing functions according to the sensors, wherein each of the predetermined error-average distributing functions is an error-average function between simulating obstacle observing datasets of a simulating obstacle and real datasets of the simulating obstacle;

providing a plurality of predetermined existence-probability datasets according to the sensors, wherein each of the predetermined existence-probability datasets is an existing probability of a signal detected by one of the sensors;

using the processor to fuse the observing positions of the obstacle and the observing velocities of the obstacle, the predicting positions of the obstacle and the predicting velocities of the obstacle and the preliminary error-average distributing function to generate a plurality of error variations and a plurality of fused obstacle datasets;

using the processor to correct an accumulation of the error variations according to the predetermined existence-probability datasets and making a judgement on whether or not the obstacle exists, wherein the processor stores a predetermined threshold value, the processor making the judgement on whether or not the obstacle exist by comparing the accumulation and the threshold value and when the accumulation is smaller than or equal to the threshold value, the obstacle is deemed to exist and when the accumulation is larger than the threshold value, the obstacle is deemed to be absent; and introducing a braking signal or an auto emergency brake according to the judgement that the obstacle exists.

2. The method of claim 1 , wherein a Kalman Filter is used in the fusing method, and each of the fused obstacle datasets comprises a fused obstacle position, a fused obstacle velocity and a fused obstacle type.

3. The method of claim 1 , wherein one of the sensors is a radar and another one of the sensors is a camera.

4. The method of claim 1 , wherein the predetermined error-average distributing functions are prepared beforehand by:

disposing a dynamical positioning module on the simulating obstacle which simulates the obstacle;

driving the dynamical positioning module to generate a plurality of dynamical positions and using the sensors to get a plurality of simulating observing positions of the simulating obstacle and a plurality of simulating observing velocities of the simulating obstacle corresponding to the dynamical positions; and

using the processor for receiving the dynamical positions and the simulating observing positions of the simulating obstacle to calculate differences thereof for generating the predetermined error-average distributing functions;

wherein each of the predetermined error-average distributing functions is corresponding to each of the simulating observing velocities of the simulating obstacle, one of the sensors comprising a field of view, and the dynamical positions and the simulating observing positions of the simulating obstacle are located inside the field of view.

5. A method for analyzing an error and an existence probability of a multi-sensor fusion, which is for detecting an obstacle on a travel direction of a vehicle, the method comprising:

providing a plurality of sensors to detect the obstacle to generate a plurality of obstacle observing datasets, wherein each of the obstacle observing datasets comprises a observing position and a observing velocity;

providing a processor to generate a plurality of predicting positions of the obstacle and a plurality of predicting velocities of the obstacle according to the observing positions of the obstacle and the observing velocities of the obstacle, respectively;

providing a plurality of predetermined error-average distributing functions according to the sensors, wherein the predetermined error-average distributing functions are prepared beforehand by:

disposing a dynamical positioning module on a simulating obstacle which simulates the obstacle;

driving the dynamical positioning module to generate a plurality of dynamical positions and using the sensors to get a plurality of simulating observing positions and a plurality of simulating observing velocities of the simulating obstacle corresponding to the dynamical positions; and

using the processor to receive the dynamical positions and the simulating observing positions of the simulating obstacle to calculate differences thereof for generating the predetermined error-average distributing functions;

using the processor to fuse the observing positions of the obstacle and the observing velocities of the obstacle, the predicting positions of the obstacle and the predicting velocities of the obstacle and the preliminary error-average distributing function to generate a plurality of fused obstacle datasets and a plurality of error variations;

using the processor to correct an accumulation of the error variations according to a predetermined existence-probability datasets and making a judgement on whether or not the obstacle exists, and introducing a braking signal or an auto emergency brake according to the judgement that the obstacle exists;

wherein each of the predetermined error-average distributing functions is corresponding to each of the simulating observing velocities of the simulating obstacle, one of the sensors comprising a field of view, and the dynamical positions and the simulating observing positions of the simulating obstacle are located inside the field of view.

6. The method of claim 5 , wherein in Kalman Filter is used in the fusing method, and each of the fused obstacle datasets comprises a fused obstacle position, a fused obstacle velocity and a fused obstacle type.

7. The method of claim 5 , wherein one of the sensors is a radar and another one of the sensors is a camera.