Ecovacs Robotics presents AI research paper at IROS 2019

20 December, 2019

Cutting-edge technology

Ecovacs Robotics was also the only company at IROS invited to share its AI research and cutting-edge technology with universities and research institutions including: the University of Maryland, the US Military Academy at West Point, and the University of Surrey in a ‘Thought Leadership’ series of documentaries produced by IROS TV.

The 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2019) is an international conference in robotics and intelligent systems, which was held in Macau from 4 to 8 November. The conference is co-sponsored by the IEEE, the IEEE Robotics and Automation Society (RAS), the Robotics Society of Japan (RSJ), and other technology associations.

Currently, 2D based VSLAM and LDS SLAM are the two most common indoor mapping and navigation technologies and Ecovacs Robotics claims to be the only home service robotics company providing robots with both VSLAM (visual-based navigation) and LDS SLAM (laser-based navigation) technologies. However, traditional VSLAM technology can be time consuming since it relies on online optimisation to provide iterative solutions and localisation.

Deep learning

Thus, deep learning has been introduced to replace online optimisation with the more efficient offline training. After the training, the robot uses the deep network for localisation and relatively shortens the time needed for optimisation. But most of the deep learning requires supervised learning with vast amounts of labelled data. The semi-supervised learning technology proposed by Ecovacs Robotics therefore solves this problem to a certain extent. The company's SLAM application is claimed to be able to improve localisation accuracy in a complex environment, enhance the performance of robots, boosts the value for users, and truly frees users’ hands.

The AI Institute (Nanjing), Ecovacs Robotics was established in July 2018 with a focus on conducting cutting-edge AI research and development, to enable the company to develop innovative new intelligent robotic solutions. In 2019, a R&D team of nearly 30 people set up by the institute successfully completed the initial construction of a high-performance GPU computing platform, distributed storage platform and mobile robot experimental platform. The team also carried out a series of researches on 3D visual perception, tactile perception, behavior recognition, and human-computer interaction to pave the way for Ecovacs Robotics' next generation of robots.

According to Ecovacs Robotics' 2019 Mid-Year Financial Report, in mainland China, the company's robotic vacuum cleaners accounted for as much as 48% of total market share.