16 March 2020 in IEEE Sensors Journal
A novel autonomous initial alignment for the stationary strapdown inertial navigation system (SINS) with the bio-inspired polarized skylight sensors (PSS) is proposed to improve the precision and convergence speed in this paper. A pair of perpendicularly polarized skylight sensors with six-channels are used to measure and calculate the polarization azimuth angles, which is finally constructed a new polarized measurement error equation. With the SINS error equations, the SINS/PSS integrated navigation error equations are developed to achieve autonomous and fast initial alignment. Two types of observability, geometric observability and stochastic observability, are introduced to evaluate the integrated navigation systems. The geometric observability, which is defined by calculating the rank of the observability matrix, is analysed. It is found that the integrated navigation systems are not completely observable. Further, the degree of observability based on singular value decomposition is also used to examine the observable degree of the SINS/PSS integrated navigation system. The stochastic observability is associated with observability analysis with the estimated precision of the designed filter. The Kalman filter is designed to estimate the unknown state based on the measurements from the inertial measurement unit and PSS. Simulations and experiments demonstrate that the SINS/PSS achieve a rapid initial alignment with higher precision. Compared with the SINS, the vertical misalignment angle accuracy of SINS/PSS is improved by more than 90%, and the time cost is shortened by 60% in the simulation test. For the initial alignment experiment, the vertical misalignment greatly improved the accuracy and shorten the time cost.
1 January 2019 in Mathematical Foundations of Computing
The self-driving based on deep reinforcement learning, as the most important application of artificial intelligence, has become a popular topic. Most of the current self-driving methods focus on how to directly learn end-to-end self-driving control strategy from the raw sensory data. Essentially, this control strategy can be considered as a mapping between images and driving behavior, which usually faces a problem of low generalization ability. To improve the generalization ability for the driving behavior, the reinforcement learning method requires extrinsic reward from the real environment, which may damage the car. In order to obtain a good generalization ability in safety, a virtual simulation environment that can be constructed different driving scene is designed by Unity. A theoretical model is established and analyzed in the virtual simulation environment, and it is trained by double Deep Q-network. Then, the trained model is migrated to a scale car in real world. This process is also called a sim2real method. The sim2real training method efficiently handles these two problems. The simulations and experiments are carried out to evaluate the performance and effectiveness of the proposed algorithm. Finally, it is demonstrated that the scale car in real world obtains the capability for autonomous driving.