[Autonomous Robot Path Planning]
[Autonomous Robot Path Planning]
Embedded Files
In this project we tried to find an intelligent navigation system in a dynamic environment for robots
In this project we tried to find an intelligent navigation system in a dynamic environment for robots
Autonomous robots have to make intelligent decisions as well as plan optimally to attain targets depending on their tasks. Moreover, the bot needs to plan a collision-free path to reduce time, energy and distance in a given environment. In search of a collision-free path planning problem faces in many disciplines such as chip design, tracking accomplishment, etc, but major in-disciplinary for autonomous bot’s navigation. Motion planning for autonomous bots is decomposed into path planning and trajectory planning. Path planning intention is to provoke a collision-free and optimized path concerning some criteria. Trajectory planning is to schedule the movement of bot along the planned path.
This work has been assigned to my PhD student. Initial Grid level path planning using heuristic approach and derivative based approach has been developed. The work has been published in ANTS 2020 Special issue.
This work has been assigned to my PhD student. Initial Grid level path planning using heuristic approach and derivative based approach has been developed. The work has been published in ANTS 2020 Special issue.
[Generating Deep Learning Networks using Swarm Optimisation Techniques]
[Generating Deep Learning Networks using Swarm Optimisation Techniques]
In this Project, we tried to identify the best deep learning network architecture for any computer vision applications with the help of swarm optimisation techniques
In this Project, we tried to identify the best deep learning network architecture for any computer vision applications with the help of swarm optimisation techniques
There have been a lot of important advancements with Deep Learning Architectures (CNN). CNN has come a long way from 5 layered LeNet to DenseNet with about 100-200 layers. There have been new and different types of layer invented like the inception layer and residual layers. Till now there have been no major advancements regarding automating the process of creating the CNN architecture. In this research work, the process of creating or selecting the Architecture of CNN is automated and the best architecture is selected in a solution space using Swarm Intelligence algorithms.
There have been a lot of important advancements with Deep Learning Architectures (CNN). CNN has come a long way from 5 layered LeNet to DenseNet with about 100-200 layers. There have been new and different types of layer invented like the inception layer and residual layers. Till now there have been no major advancements regarding automating the process of creating the CNN architecture. In this research work, the process of creating or selecting the Architecture of CNN is automated and the best architecture is selected in a solution space using Swarm Intelligence algorithms.
Current Status
Current Status
This work has been carried out by me with the help of Bachelor Students in SRMIST. This work has passed its half milestone and We have generated the best architectures with respect to the data is found. The final interface and checking with huge datasets is remaining as a part of final testing. The work has been communicated for the journal of artificial intelligence..
This work has been carried out by me with the help of Bachelor Students in SRMIST. This work has passed its half milestone and We have generated the best architectures with respect to the data is found. The final interface and checking with huge datasets is remaining as a part of final testing. The work has been communicated for the journal of artificial intelligence..
Page updated
Google Sites
Report abuse