Search

This paper proposes a method to realize the robot’s self-learning of environment by using an episode cognition model inspired by the hippocampus. The episode cognition map is suitable for robot navigation in an unknown environment, which solves the problem of robustness of the robot’s perception in complex and dynamic environments. The model called the hippocampus episode cognitive network (HEC), is based on the hippocampus CA1, CA3, and the dentate gyrus, and combined with the adaptive resonance theory (ART) network. It extract new events through the incremental generation of cognitive neurons, and encode the events into episodes through space-time connections. The episode nodes can be connected to generate an episode cognition map. This method can learn, store, and update information for autonomous mobile robots in an unknown environment. Based on the episode cognitive map, the path trajectory can be predicted through the playback of the episode neuron. The experimental results on the mobile robot show that this method can effectively improve the robot’s adaptability for location and mapping in a complex and dynamic environment.

The traditional Cubature Kalman Filter (CKF) and its derived algorithms cannot work without the two hypotheses of Kalman Filter (KF), one is that the system model is accurate and the other is the system is only influenced by independent white noise with known statistical characteristics. However, it is difficult to fully guarantee the above hypotheses in actual operation. The presence of multiplicative noise undermines the first hypothesis, while the additive noise correlation undermines the second hypothesis. Given invalid hypotheses, the estimated performances of CKF and its derivative algorithms are degraded drastically. To solve this issue, this article proposes Multiplicative Noises and Additive Correlated Noises Cubature Kalman Filter (MACNCKF), which solves state estimation problems involving multiplicative noise and additive noise while maintaining CKF advantages. Moreover, this algorithm has been testified to be correct and easy to be transplanted to CKF and its derivative algorithm after strict mathematical derivation. When the system lacks multiplicative noise and additive noise, the algorithm is degraded to corresponding algorithm in a general form, which helps extend the application environment of CKF and its derivative algorithms, thus improving robustness. Numerical simulation and experiments on quadruped robot system indicate that MACNCKF can effectively solve state estimation problem involving both multiplicative noise and additive noise. Given time consumption in MACNCKF basically comparable to that in CKF, MACNCKF has improved estimation accuracy, robustness and reliability.

Robots manipulating in domestic environments generally need to interact with articulated objects, such as doors, drawers, laptops and swivel chairs. The rigid bodies that make up these objects are connected by a revolute pair or a prismatic pair. Robots are expected to learn and understand the objects' articulated constraints with a simple interaction method. In this way, the autonomy of robot manipulation will be greatly improved in an environment with unstructured constraints. In this paper, a method is proposed to obtain the articulated objects' constraint model by learning from a one-shot continuous visual demonstration which contains multistep movements, and this enables human teacher to continuously demonstrate several tasks at once without manual segmentation. At the end of this paper, a six-degree-of-freedom robot uses the constraint model obtained by demonstration learning to achieve manipulation planning of various tasks based on the AG-CBiRRT algorithm.