Reduce the on-site deployment time of commercial robots by 80-90%. Is it a gimmick or real power?
Time-consuming and laborious on-site deployment
Unlike sweeping robots that are ready to use upon startup, commercial robots generally have a final "process" of on-site deployment before use. This process requires dedicated on-site deployment engineers to perform professional operations such as SLAM mapping, target point annotation, and testing area by area. In addition to subsequent training, the entire process is cumbersome and takes a certain amount of time.
Theoretically, if the approval process goes smoothly (some locations require special licenses for installation), the elevator control installation and scanning are synchronized, and the functional test is successful, the deployment of commercial robots can take up to 1.5 working days at the earliest. However, in the real environment, the deployment time of robots in different landing scenarios is quite different. Due to the technical limitations of traditional marker positioning and laser positioning, in shopping malls, supermarkets and other scenes with large spaces, high ceilings, and complex light environments, there are The cost of labeling is high, the space is difficult to label, the space layout needs to be modified, etc., and it is prone to inaccurate positioning due to environmental changes, thus greatly extending the deployment time. In addition, due to this technical limitation, when the scene layout is changed later, the robot manufacturer must be contacted to arrange for engineers to arrive on site to re-map the robot and plan the route, which not only increases the end user's usage costs, but also further increases the manufacturer's operation and maintenance costs.
In the early days, when the number of robots implemented was small, the more cumbersome deployment process had no real impact. However, as the market continues to develop, the deployment efficiency of robots has obviously failed to keep up with the speed of shipments, resulting in rising costs and long delivery cycles. , an increase in negative reviews and many other problems began to emerge.
The pain points are difficult to solve. How to upgrade the deployment plan?
When laser technology fails, visual technology becomes the new choice. For example, Purdue has shifted its technical route to VSLAM and applied it to the "Happy Send 2" robot. This robot does not need to be coded during the deployment process, and has higher stability and site adaptability. Compared with most market solutions, it reduces the cost by 75%. deployment time. Coincidentally, INDEMIND, as one of the first AI technology companies in China to deploy computer vision, is also at the forefront of the industry in visual research and development. The "Commercial Robot AI Kit" visual fusion navigation solution it launched also has the same advantages.
"Commercial Robot AI Kit" is based on INDEMIND's self-developed INDEMIND OS Fusion AGI system. It adopts a multi-sensor fusion architecture with a binocular stereo camera as the core, supports different categories of mainstream sensors on the market, and meets the needs of commercial robot navigation, positioning, and intelligence. The development of core functions such as obstacle avoidance, path planning, and decision-making interaction can be widely used in commercial service robot platforms such as commercial cleaning, hotel distribution, food delivery, and inspection.
Robots equipped with INDEMIND's "Commercial Robot AI Kit" do not need to be coded, and the overall on-site deployment time can be reduced by 80-90%. New equipment and new scenarios can already be used immediately like a sweeping robot.
First of all, after the new equipment enters the scene, the user can directly push the robot to plan any operation area or operation path. After completion, the robot can complete the operation planning independently.
After extensive adjustments are made to the scene, the robot can autonomously detect scene changes during operation and update the map in real time, without the need for engineers to participate, to achieve stable and continuous operation.
In terms of technical implementation, based on the industry-leading embedded deep learning and VSLAM algorithms, the robot can quickly output individual object semantics and regional scene semantics. The robot can independently construct a 3D semantic map and high-precision positioning of very large scenes in real time, and in the process , the system can assist in real-time annotation of points such as elevators, turnstiles, and narrow aisles. At the same time, in conjunction with a set of visual deployment tools developed by INDEMIND that simplify and facilitate operation, novice users can complete autonomous operations through the machine + the LCD screen configured on the machine side without the involvement of engineers, which is simple and fast.
In addition, INDEMIND has specially built a robot deployment simulation platform. Before the robot is provided to users, it will support users to perform deployment simulations based on this simulation platform to verify that it fully meets user needs before delivery, greatly reducing the difficulty of deployment and improving deployment efficiency. .
It is worth mentioning that INDEMIND has also developed a systematic environmental fill-in strategy, including active ambient fill-light configuration and mapping strategies under changing lighting conditions. In actual performance, in the face of direct strong light, no light source, It can work without any difference in dim and other special lighting environments, meeting all-weather operation requirements.
At present, INDEMIND has reached cooperation with many giant customers at home and abroad. Among them, the medium-sized cleaning robot Cobi18, which is cooperated with ICE, a traditional cleaning equipment manufacturer, has been deployed in batches in more than a dozen countries around the world and has achieved zero-fault operation in the European and American markets. In the future, INDEMIND's expected orders in three years will exceed 100,000 units.