Idea Name: |
DeepSLAM |
Slogan: |
Gotta go deep |
Supervisor Name: |
Jawad Khan |
Supervisor Designation: |
Assistant Professor |
Supervisor School: |
SMME |
Supervisor Department: |
Department of Robotics and Artificial Intelligence |
Contact number: |
03455496789 |
Email ID: |
[email protected] |
Abstract: |
The rise of Increasing Planetary exploration in particular on the Martian and Lunar surfaces have given rise to new problems in Autonomous systems. Mapping and Navigating is not an easy task in unstructured planetary environment like that on the Moon, in order to aid existing algorithms, we propose |
What is the unmet need in society that your idea will fulfill ? |
Provide better autonomy for rovers involved in Planetary exploration and mining. In the future this will aid in reducing the materialistic and environmental burden on sources on earth, provide new sources for metals like Lead and Tin running out and help turn humanity into an inter-stellar species. |
Who needs it ? How many would benefit ? |
Many space agencies and research organizations like ISAS, DLR and NASA are working on SLAM for Unstructured Planetary environments. This research will eventually benefit such organizations and ultimately the entire space sector, aiding mankind in its exploration of the cosmos. |
How will the solution works |
Our Solution involves the use of Neural networks for performing SLAM alongside algorithms. Neural networks will aid in many areas like Loop Closure Detection, pose estimation, Sensor fusion and filling in occluded data. Our Neural networks will be trained in simulation environments, in case they show superior performance to algorithms, they will be implemented in favor of them after which we will present our result as a ROS Implementation. |
Who are your competitors ? How is your solution different |
We have a lot of competitors from both Industry and Academia. Neural networks have been used previously in SLAM, however mostly in Urban environments. What makes our solution different is the use of multiple neural networks for different aspects of SLAM in remote environments. |
Status: |
new |
Entry Date & Time: |
2022-12-15 (0706) |