Smart security robot can be used in various scenarios, including enclosed environment for inspection robot, like computer rooms, e-commerce warehouses and nuclear power plants, as well as open space for patrol robots, like CKTs, tracks, parks and public places.
Compared with patrol robots, inspection robots have higher requirements for navigation accuracy and professional testing ability, and they often use basic AI functions like smoke monitoring, high temperature warning and strange sound alert.
Patrol robot in the contrast enjoys a larger demand in the market for it can be used in rich scenarios. As they usually work in the outdoor environment, patrol robots have higher requirement for navigation stableness, functioning smoothness and robustness of itself, and the main AI functions involved are facial recognition, pedestrian detection, vehicle recognition and testing, smoke recognition, strange behavior analysis, person-certificate verification, speech interaction and semantic analysis.
Inspection, patrol and monitoring are the three pillars in the security industry, and smart security robot furthered the development of active security safeguard. The core technology of smart security robots is low-speed autonomous driving technology, which is composed of chassis technology, sensor packages and auto-pilot SLAM technology, AI technology and network transmission technology based on computer vision and cloud platform management & control technologies.
There are two tricky issues in applying smart security robots in real life. One is to guarantee the stableness and accuracy of SLAM in dynamic navigation. The other is to ensure the availability and accuracy of AI technology in complex scenarios, such as scenarios with diversified intensity of illumination, bad weather or multiple moving subjects.
Currently, smart security robots are not fully intelligent in self-learning, self-management and self-decision making, while it has possessed various basic AI functions, such as target detection, item recognition, environment perception and multi-mode interaction. Future breakthroughs in related technologies rely on real-life application of academic findings as well as real demands in the market.
More details can be seen in the full version of 2019 China's AI + Security and Protection Industry Report