AI+VSLAM visual technology revolution, Rockchip helps in the industrial upgrading of the fourth-generation sweeping robots
In recent years, the sweeping robot industry has developed rapidly. After the three modes generations of random, planning and navigation, it still requires much manual intervention since the difficult problems such as interactive modes, intelligent planning and hardware energy consumption remain unsolved. Rockchip has released four Linux-based integrated chip-level solutions, which can help in the industrial upgrading and effectively solve the difficult product problems.
The flagship RK3399 solution adopts AI+VSLAM positioning and navigation technology, RV1108 and RK3326 support VSLAM visual positioning and navigation technology, and RK3308 supports the mainstream laser navigation technology.
AI + visual recognition + VSLAM positioning and navigation technology
Rockchip has launched the AI + VSLAM positioning and navigation technology based on RK3399, which enables the sweeping robot navigation technology to enter the fourth-generation directly, truly realizing AI, and with the image recognition, big data, deep learning and intelligent algorithm, perfectly integrating artificial intelligence and sweeping robots.
Adopting the fourth-generation navigation technology, Rockchip RK3399 breaks through the limitation of the reliance of the original VSLAM technology on light conditions, based on visual recognition technology and AI algorithm. It can not only be applied to complicated indoor scenarios such as weak light, weak texture and strong light, but also intelligently identifies nearly a hundred indoor image objects, classifies records, performs visual ranging, avoids obstacles and plans the best sweeping path.
Mainstream solutions: RV1108, RK3326 and RK3308
Features of RV1108 and RK3326:
VSLAM visual positioning and navigation technology
Features of RK3308:
Laser SLAM positioning and navigation technology
The RV1108 and RK3326 integrated solutions issued this time are both equipped with VSLAM visual positioning and navigation technology. Although these two major navigation technologies are already matured, Rockchip’s brand new solutions are more advantaged in hardware performance. To satisfy the R&D demands of different customers, the two types of solutions differentiate the implementation of the VSLAM algorithm:
RV1108 supports the implementation of VSLAM algorithm on DSP and the built-in 8M ISP can simultaneously support 3D cameras. For customers using the VSLAM positioning and navigation technology, it can adapt to more complicated application scenarios; it supports the MIPI CSI interface with a wider range of Camera Sensor module selections.
RK3326 can support the technical implementation of VSLAM algorithm on CPU/GPU. RK3326 GPU adopts G31, which can support Full profile OpenCL2.0, allowing fast implementation of VSLAM algorithm on GPU. The concerted work of CPU + GPU brings better performance.
RK3308 can support the laser navigation technology of the mainstream sweeping robots. Compared with the 40nm Cortex A7 solutions adopted by its peers, with the 64-bit Cortex A35 structure, all the Rockchip solutions use the 28nm new processor with higher performance and lower power consumption, effectively solving the chip heating problem, and this is particularly important for sweepers with built-in batteries.
Five major technical advantages of Rockchip fourth-generation sweeping robot
In terms of technology, Rockchip’s four major solutions have the following advantages over traditional sweeping robot solutions:
5. Customized design of flexible hardware solutions, higher cost performance
Innovation and flexible design of hardware
Rockchip has many algorithm partners and ODM/OEM customers in sweeping robot ecology, covering laser, VSLAM, AI, etc. and is able to achieve rapid mass production. In response to the different product innovation needs of industrial customers, Rockchip has also carried out the flexible design of hardware for the four major sweeping robot solutions. On peripheral materials and parts, RK3399, RV1108, RK3326 and RK3308 may provide a broader support list for such key components as DDR/WIFI/BT/Camera/Nand Flash/EMMC. For partners, it will substantially improve their market competitiveness.
In the fast growing industry, by virtue of the experience in AI, image vision and chip technology, Rockchip’s first-released solutions are very overturning compared with traditional products. The integration with the brand new IoT and AI will enable the white goods market to secure more opportunities for breakthroughs among limited changes.
Home appliances with embedded intelligent technologies become popular. Another general trend is AI home appliances based on the users’ life and business scenarios, which is also in line with the era characteristics of AI and IoT.
Pre：AI Computation Performance Improves and Rockchip Officially Launches Android 8.1 NNAPI SDK Next：Rockchip released the deep learning-based target detection solution, speeding up the commercial process of high-end AI