In the field of environmental recognition, advances are being made in the use of inexpensive vision cameras and depth cameras, and in understanding the meaning of acquired data.In the future, it is expected that navigation systems that can cope with unknown environments and environments with dynamic changes will be developed. In terms of control, integrated control of manipulation and autonomous mobile robots is expected.
A comprehensive action plan for research and development in the field of robots and social implementation (robot action plan)
Direct Visual SLAM (Kudan)
By combining GN-Net feature extraction and Visual SLAM, we perform semantic peripheral recognition and realize self-position estimation that is robust to dynamic changes.
Consideration of depth cameras and depth cameras is also seen (mid 2010s)
2010 Heihagi Scale 3D SLAM using LILJAK 2010 Hachishunkan The use of vision cameras has been increasing. For indoor use, the use of fat cameras has been attracting attention. There are also technologies that achieve highly accurate environmental recognition by integrating multiple types of sensor information or combining neural networks. Kudan made SLAM robust to dynamic changes by combining it with semantic understanding of the scene. Yonetani et al. developed a machine learning-paced A* search algorithm and succeeded in generating an optimal path plan even for unknown rings.
The first action plan in the mobility area is to respond to unknown and dynamically changing environments such as outdoors.
development of robust autonomous mobility technology. It is necessary to develop algorithms that allow robots to operate stably and autonomously even in unknown environments where there is no prior map information or in environments where dynamic changes occur.It is necessary to aim to simplify and eliminate the need for advance map creation and updating. Taken as a reference example, Yonetani et al. developed a
Technology development trends - Mobility
In the field of environmental recognition, advances are being made in the use of inexpensive vision cameras and depth cameras, and in understanding the meaning of acquired data.In the future, it is expected that navigation systems that can cope with unknown environments and environments with dynamic changes will be developed. In terms of control, integrated control of manipulation and autonomous mobile robots is expected.
A comprehensive action plan for research and development in the field of robots and social implementation (robot action plan)
Direct Visual SLAM (Kudan)
By combining GN-Net feature extraction and Visual SLAM, we perform semantic peripheral recognition and realize self-position estimation that is robust to dynamic changes.
Consideration of depth cameras and depth cameras is also seen (mid 2010s)
2010 Heihagi Scale 3D SLAM using LILJAK 2010 Hachishunkan The use of vision cameras has been increasing. For indoor use, the use of fat cameras has been attracting attention. There are also technologies that achieve highly accurate environmental recognition by integrating multiple types of sensor information or combining neural networks. Kudan made SLAM robust to dynamic changes by combining it with semantic understanding of the scene. Yonetani et al. developed a machine learning-paced A* search algorithm and succeeded in generating an optimal path plan even for unknown rings.
The first action plan in the mobility area is to respond to unknown and dynamically changing environments such as outdoors.
development of robust autonomous mobility technology. It is necessary to develop algorithms that allow robots to operate stably and autonomously even in unknown environments where there is no prior map information or in environments where dynamic changes occur.It is necessary to aim to simplify and eliminate the need for advance map creation and updating. Taken as a reference example, Yonetani et al. developed a