Adaptive Multi-Sensor Fusion In the real world, it is difficult to obtain accurate information about the environment, and it is impossible to ensure that the sensors can always work properly. Therefore, for various uncertain situations,

Robust fusion algorithms are very necessary. Some adaptive multi-sensor fusion algorithms have been developed to deal with the uncertainty caused by the imperfection of sensors. For example, Hong proposed an extended joint method through innovative technology.

It is possible to estimate the optimal Kalman gain for filtering a single measurement sequence. Pacini and Kosko also developed an adaptive target tracking fuzzy system that can be used in mild environmental noise.

It incorporates the Kalman filter algorithm in the processing process.

Navigation and positioning

In robotic systems, autonomous navigation is a core technology and a key and difficult issue in the field of robotics research. The basic tasks of navigation are three: (1) Global positioning based on environmental understanding:

By understanding the scenery in the environment, identifying artificial landmarks or specific objects, the robot can be positioned and provide materials for path planning; (2) Target recognition and obstacle detection:

Detect and identify obstacles or specific targets in real time to improve the stability of the control system; (3) Safety protection: It can analyze obstacles and moving objects in the robot's working environment and avoid damage to the robot

。

The robot has a variety of navigation methods, which vary according to factors such as the completeness of environmental information and the type of navigation signal.

, can be divided into three categories: map-based navigation, map-based navigation and map-free navigation. According to the different hardware used for navigation, the navigation system can be divided into visual navigation and non-visual sensor combined navigation [8]

Visual navigation uses cameras to detect and identify the environment to obtain most of the information in the scene. The content of visual navigation information processing mainly includes: visual information compression and filtering, road surface detection and obstacle detection.

, recognition of specific environmental signs, and three-dimensional information perception and processing. Non-visual sensor navigation refers to the use of multiple sensors working together, such as probe type, capacitive type, inductive type,

, mechanical sensors, radar sensors, photoelectric sensors, etc., are used to detect the environment and monitor the robot's position, posture, speed, and internal state of the system.

Perceiving the static and dynamic information of the robot's working environment enables the robot's corresponding work sequence and operation content to naturally adapt to changes in the working environment and effectively obtain internal and external information.

In autonomous mobile robot navigation, whether it is local real-time obstacle avoidance or global planning, it is necessary to accurately know the current state and position of the robot or obstacles to complete navigation.

, obstacle avoidance and path planning, etc., this is the robot's positioning problem

The more mature positioning systems can be divided into passive sensor systems and active sensor systems. The passive sensor system senses the robot's own motion state through encoders, acceleration sensors, gyroscopes, Doppler velocity sensors, etc.

The positioning information is obtained through cumulative calculation. The active sensor system uses active sensors such as ultrasonic sensors, infrared sensors, laser rangefinders, and video cameras to perceive the robot's external environment or artificially set road signs.

Matching with the system's pre-set model can obtain the relative position of the current robot and the environment or landmarks, and obtain positioning information.