“Find: with the continuous release of market demand, the shipment of robots is growing. However, from concept to field application, the landing process of robots is not smooth. Among them, the field deployment is a common problem that has been puzzling manufacturers and end users.”
As the final link of robot scene application, on-site deployment seems to have no complicated technical details from an intuitive point of view. It is just on-site debugging, which does not take much time. However, is this really the case?
Take the cleaning robot as an example:
- After the robot is delivered to the customer by the manufacturer, the engineer starts to deploy the network, elevator control and program-controlled telephone system;
- Then engineers began to push the robot to build maps one by one;
- After the completion of the drawing under construction, the engineer will spend some time to revise the map;
- Then the engineer began to mark and take positions for different target points one by one and special treatment, such as waiting points in narrow areas such as corridors; Ramp, ridge and other deceleration points; Glass, escalator and other dangerous points; Gate, elevator and other special points often require a lot of time during this period;
- After marking, the final test, training and delivery shall be carried out.
The above is the standard process for the deployment of the vast majority of robots in the market, that is, the deployment scheme of “robot mapping based business management based on management tools”. Due to the complexity of the process, it usually takes a long time to deploy. From a technical point of view, this scheme has many defects:
- The robot deployment operation process is complex and difficult (such as slam mapping, target point marking, etc.), and all deployment operations can only be performed by the on-site deployment engineers specially trained by the robot company, which is difficult for users to operate and participate;
- The on-site deployment engineer is required to implement all changes, so the user cannot operate independently and the environmental adaptability is poor;
- Robot companies need on-site deployment engineers for on-site operation, and the traffic cost is high. Many companies have even established deployment teams in major cities, which is very expensive for companies;
- Before the final mass production of the robot, most manufacturers will cut the cost in multiple rounds, making the performance redundancy of the robot low. Once faced with more complex scenes, it often needs to be debugged many times, build maps repeatedly, and increase the deployment time.
In the actual field deployment process, the low deployment efficiency will lead to the increase of investment cost and time cost for both manufacturers and end users. Therefore, with the continuous release of market demand, it has become a practical task to solve the problem of deployment efficiency.
How to solve this problem? Can deploying service robots be as simple and fast as using household sweeping robots?
In fact, this idea has been realized by a company. Relying on years of self-research experience in robot full stack technology, index has developed a new robot deployment scheme.
Different from the market scheme of “robot based mapping and business management based on management tools”, the index scheme is a way of “robot based mapping and business management”. It is equipped with an independently built simulation and verification platform, deployment process and management tools, which can minimize the participation of technicians and realize users’ independent deployment.
In terms of application, the index robot deployment scheme has five technical features:
Real time 3D semantic map construction and system aided annotation
In order to simplify the deployment process, based on the industry-leading binocular stereo vision technology, combined with the embedded deep learning and vSLAM algorithm at the device end, index can quickly output the semantics of individual objects and regional scenes, and can realize the real-time construction of 3D semantic maps and high-precision positioning. When marking some target points such as elevators, gates and narrow aisles, the system can assist in marking in real time based on object recognition, Reduce the difficulty of operation, and ordinary users can execute independently without the participation of engineers.
Simple machine operation
Index has developed a set of visual deployment tools for simple and convenient operation, which can enable users to realize independent operation based on the LCD screen configured on the machine + machine side, which is simple and convenient.
– robust security strategy
Based on the fusion method of multiple sensors such as binocular and lidar, and combined with in-depth learning, index has developed a set of decision engine technology to greatly improve the safety decision-making level and realize the safety management of escalators, glass, falls and other scenes.
Cloud status control
Index has specially built an intelligent management and control platform in the cloud. Through the powerful computing power deployed in the cloud, it provides real-time full state management and control intelligent services for robots. Whether domestic robots or commercial robots, it greatly improves task processing and problem response capabilities without increasing the front-end hardware cost.
Simulation verification platform
Index has its own robot deployment simulation platform. Before the robot is provided to the user, it will support the user to carry out deployment simulation based on the simulation platform, so as to verify that it fully meets the user’s use needs before delivery, greatly reduce the deployment difficulty and improve the deployment efficiency.
By solving the process problems in the field deployment one by one, the index robot deployment scheme is not only the improvement of deployment efficiency, but also the change iteration of new and old technologies. Under the background of the continuous expansion of market scale and the continuous growth of shipment, the application of the new scheme undoubtedly has practical value for manufacturers, end users and the overall development of the robot industry.
Reviewed and edited by: Fu Qianjiang