Development of low cost 3D obstacles detector for limited/reduced velocity AGVs

• Grant number: 17/08692-0
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Duration: May 01, 2018 - January 31, 2019
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Patrick Yuri Shinzato
• Grantee: Patrick Yuri Shinzato
• Host Company: 3DSoft Consultoria Ltda. - ME
• CNAE: Fabricação de equipamentos e aparelhos elétricos não especificados anteriormente
• City: São Carlos
• Associated scholarship(s): 18/11993-4 - Development of low cost 3D obstacles detector for limited/reduced velocity AGVs, BP.TT; 18/10960-5 - Development of low cost 3D obstacles detector for limited/reduced velocity AGVs, BP.PIPE

Abstract

Perception system in robots is essential to enable a high-level of autonomy. The growing inclusion of robotic system to perform manual tasks have triggered a vast demand to sensors, such as cameras and LiDARs. However, one of the biggest barriers to develop autonomous robots is the high cost of the sensors and can correspond to most of the final product. Despite the existence of low cost sensors, they lack in the precision and/or resolution and make them unfeasible to use. Another reason that difficult the access to the sensors is the dependence on third party modules to interpret the sensor data. In this sense, it still needs a big technological advance to popularize these sensors. There are initiatives from international companies proposing the development of low cost sensors, but their price is still prohibitive when considered the commercialization in the Brazilian market. Moreover, there are few companies researching the use of embedded computers for integrated processing of the sensor data. This project proposes the development of a low cost sensor that combines methods to enhance the quality of the captured data and a 3D obstacle detection method. The proposed product will be formed by a pair of cameras to generate stereoscopic images and a single beam LiDAR to enhance these images through sensor fusion. All processing will be performed in an embedded computer that is also responsible for obstacle detection. This sensor has a high commercial appeal, mainly in companies that require autonomous vehicles (AGVs - Autonomous Ground Vehicles) that travels in low speed and in agricultural robots that operates in outdoor environments. (AU)