en.Wedoany.com Reported - Norinco's Hangzhou Zhiyuan Research Institute has successfully developed an intelligent inspection robot dog, achieving autonomous navigation and proactive security patrols in complex campus environments without the need for pre-set maps. The device possesses three core capabilities: pure visual map-free autonomous walking, human-machine collaborative operations, and all-weather, all-scenario intelligent perception. It addresses pain points in traditional campus inspections, such as numerous blind spots, delayed response, and high labor costs.
Frequent transitions between indoor and outdoor campus scenes and dense obstacles often cause traditional navigation algorithms to fail. To address this, the research team collected massive amounts of data, adapted the algorithm to various scenarios, and iteratively optimized it, ultimately achieving pure visual map-free autonomous navigation. The intelligent inspection robot dog does not require prior surveying or mapping, nor does it need fixed routes. It can autonomously plan paths and avoid obstacles in complex environments. Its trajectory tracking error does not exceed 10 centimeters, the campus navigation success rate is no less than 90%, and its response time to dynamic obstacles is within 100 milliseconds.
To achieve deep human-machine interaction, the team integrated visual recognition and natural language understanding technologies. The intelligent inspection robot dog can identify campus personnel and understand natural language commands. Its natural language understanding response accuracy exceeds 95%, and its campus personnel recognition accuracy is no less than 85%. It can autonomously follow and conduct synchronized patrols, suitable for scenarios such as routine inspections, fault diagnosis, and on-site surveys.
Addressing the limitation of traditional surveillance systems that only record video without proactive prediction, the team, guided by real security scenarios, trained multi-target recognition and behavior analysis algorithms using massive amounts of real-world scene data. The robot dog possesses all-weather, all-scenario intelligent perception capabilities, with a risk event recognition accuracy exceeding 95%. It can autonomously identify high-risk events and push early warning information within one second, advancing the front-loading and intelligentization of security prevention and control.
Hangzhou Zhiyuan Research Institute stated that it will continue to optimize the device's autonomous decision-making and collaborative operation capabilities, expanding its application scenarios to smart campuses, industrial inspections, and emergency security.
