中国·yl23455永利(集团)有限公司-官方网站

Torchstar Partners with Nvidia and Jingdong to Launch AMR Collaborative Autonomous Mobile Machine? "Syrius Torchstar AMR.

In today's world where everyone has access to "cell phones" and "The Internet", an efficient feedback loop has been formed between producers and end-consumers, and the supply chain from producers to consumers has been drastically shortened. The rapid changes in the retail industry and the end-consumer's return to the value chain's core have made the "experience economy" increasingly important. The degree of informatization and automation in the field of warehousing and logistics is still underdeveloped. This has become the main reason to restrict the development of e-commerce and retail enterprises. The new logistics solution supported by autonomous mobile robots (Auto Mobile Robot) and artificial intelligence technology will bring a "fast" and "flexible" transformation experience, maximizing logistics potential and realizing logistics value.

On November 20, 2018, "Syrius Torchstar" (hereinafter referred to as "Torchstar") held the "2018 Syrius Torchstar Brand and Product Launch" at Hyatt Regency Suzhou. The theme of the conference was "New AI, New Logistics". At the conference, Torchstar, NVIDIA and Jingdong Logistics jointly launched a very disruptive product - "Torchstar AMR"? machine collaboration type autonomous mobile machine?

This time, the guests attending the launch mainly came from the fields of technology and investment. At the same time, authoritative media such as Xinhua and Dongfang TV were arranged to conduct interviews. As partners of Torchstar, NVIDIA and Jingdong Logistics participated in the joint launch and delivered speeches.

NVIDIA Director of Intelligent Machines, Murali Gopalakrishna delivered a joint speech, mentioning that NVIDIA has long-term cooperation with Torchstar and was surprised and appreciated that Torchstar could release new brands and products in such a short period of time. In this release, Torchstar uses one of NVIDIA's best intelligent machine platforms "NVIDIA Jetson TX2". He said that NVIDIA and Torchstar will continue to cooperate and hope that both sides will work together to promote the development of the industry in many fields such as warehousing and logistics.

During the joint launch session of Jingdong Logistics, Zhang Zu, Investment Director of Jingdong Logistics, also expressed his appreciation for the new products jointly released with Torstar. He mentioned that in this year's 11.11, Jingdong Logistics has put into use the largest robot warehouse group in China, and there are more than 50 unmanned warehouses at different levels, which provide strong support for consumers to enjoy the 11.11 Global Goodies Festival with the efficient operation of intelligent equipment. Jingdong and Torchstar teams have communicated a long time ago, and both sides also have some practical cases of cooperation, and in the establishment of Torchstar technology to determine the close cooperation in the field of warehouse logistics robot, the next two warehouses in North China and East China will cooperate, on the other hand, the two sides are also exploring the possibility of cooperation at the capital level.

The new product released by Torchstar is a robot that took half a year to design and develop before completion - "Torchstar AMR". It is a universal handling platform with reliable autonomous movement capability. It also has rich software, electrical and mechanical interfaces. It is mainly used in the field of e-commerce warehousing.

"The main features of Torchstar AMR are shown in the figure below.

It is worth noting that at the technical level, the development of "Torchstar AMR" is all system and not algorithm-oriented - although algorithms are still the core of the system but more focused on the overall safety, reliability and stability of the system. Torchstar's first AMR not only runs the leading sensor fusion algorithm, positioning algorithm and navigation algorithm but more importantly, it does not use "fetishism" to put off system problems," said Chao Jiang, founder of Torchstar. Torchstar has developed a framework based on "inter-thread communication" to help the robot interact with each other and synchronize the sensors. This is the only way to avoid many of the problems associated with ROS, the open-source software used by many other robots, such as the single point of vulnerability of a MASTER node that can cause a global system crash, and the high latency caused by packages such as TF, which reduces real-time control.

Torchstar's product also achieves the requirement of reducing the total system cost by making up for the lack of sensors with good algorithms. Chao Jiang said: The purpose of Torchstar's R&D is not to "show off" but to improve cost performance and create value for customers. Therefore, we use very simple sensors in AMR products: "single line short distance" LIDAR, "monochrome" camera, "MEMS" inertial measurement devices, etc.. At the same time, we use leading sensor fusion algorithms to synthesize the information collected by these sensors to help robot localization and map building. We have developed this system to work immediately in a wide range of environments without any prior modification to the environment.

In the "decision layer", Torchstar AMR uses many deep neural network-based algorithms, as well as a series of "traditional" algorithms such as geometric algorithms for computer vision and laser navigation algorithms.

In terms of scene implementation, Torchstar AMR products are mainly used to solve handling problems in many fields such as e-commerce warehousing. Compared with existing solutions in the market, Torchstar AMR is more flexible when working in warehouses.

According to Chao Jiang, traditional AGV systems require a central control server, and each AGV cart does every action - for example, turn left, turn right, stop, move one meter forward, etc. - and requires a central control server for "scheduling". Torchstar AMR, on the other hand, is smarter, with powerful computing, sensing and autonomous navigation capabilities. All operations during its movement are decided by each robot independently, without the need for "scheduling" by the central control server. When multiple Torchstar AMR robots form a cluster, they rely on a "distributed control" approach to collaborate with each other.

According to the report, Torchstar AMR robots, cluster management software and cloud services are all optimized for the picking process of e-commerce warehouses. Currently, Torchstar is using the robot to work with two regional branches of Jingdong Logistics in North and East China to build a prototype warehouse.

In addition, the reporter learned that the upgrade of the existing "traditional warehouse" is the first "stock" market that Torchstar wants to enter, but Torchstar is more interested in the "incremental" market of the new retail sector.

Jiang Chao said: the rise of new retail needs the support of new logistics, and new logistics needs new tools and new methods. Warehouses are rapidly entering the city. The offline experience stores of "front warehouse" and "front store and back warehouse" are becoming more and more common. These incremental markets require flexible, accurate and efficient solutions. "Torchstar AMR is a powerful and flexible efficiency tool that can better support customers in the e-commerce and retail sectors. "Syrius Torchstar serves consumers by serving the service industry. "We are committed to bringing autonomous mobile robots into various industries to optimize handling operations and accelerate industrial upgrading."