E

End_to_End ADAS

Detecting unavoidable frontal, rear-end, T-bone and other types of collisions to activate preventive measures. Monitoring the driver to keep them alert at all times.

E

End_to_End ADAS

Detecting unavoidable frontal, rear-end, T-bone and other types of collisions to activate preventive measures. Monitoring the driver to keep them alert at all times.

The problem End_to_End ADAS solves

Most numbers of deaths in the world are due to the negligence of following proper traffic rules and regulations while driving on roads. From the start to this age we made vehicles stronger but now let's make them smarter. Removing human bias while driving can lead to a better driving experience. This also opens up many more opportunities in the space of Smart Cities, Employment, etc. Leveraging the speed of today's machines which can take very quick decisions we can make the overall driving experience much safer and more enjoyable by spending time on what matters more for us than the task of spending a lot of time on the task of driving. This kind of technology can also leverage the potential of fast connections to make inter-vehicular communication networks better. Thus, smarter systems mean a better lifestyle and better lifestyle means focusing on things we care most.

Challenges we ran into

  1. The unavailability of proper documentation on the CARLA simulator meant we needed to spend a lot more time to get familiar with the simulator.
  2. Training the networks for performing detection and segmentation for vehicle localization and planning takes a lot of time due to the network architectures and model size.

Discussion