2020 is Not Yet for Level 5 Autonomy

Most major automobile companies have promised that their Level 5 autonomous vehicle will be on the road in a very near future possible 2018, 2019, and 2020. It is however 2020 already! We do not witness near level 5 autonomy in any of the autonomous vehicle companies.

I have been worried about the hype of the autonomous vehicle industry’s fully autonomous level. There have been multiple AI winters due to the same mistake made by AI researchers during the last several decades. I studied AI during two of the AI winter periods. The mistake was to promise what they cannot provide. Level 5 fully autonomy is more like an ideal concept for now at least to me. Level 5 autonomy cannot be materialized in the near future.

The article from the IEEE spectrum, Surprise! 2020 Is Not the Year for Self-Driving Car shows the Autonomous Vehicle industry will shift its focus on Level 4. Some researchers note Level 4 as the Human Centered AI.

This is not a dire expectation in the areas of autonomous driving research. This means, on the contrary, unlike nicely curated video demos from the automobile industry, there still are many challenges to be addressed by researchers.

References:

New Bimians 2020

Welcome to the BIMI Lab. You are a Bimian now.

  • Bharath Kumar Karre
  • Deepak Mysuru Heeranna
  • Meher Adheeth Hundi
  • Sanjyot Thete
  • Sumanth Pobala
  • Swathi Priya Soogoor

Development of Autonomous Vehicle Research Platform

A new project, “Development of Autonomous Vehicle Research Platform using Deep Learning and Robot Operating System” has been approved. This project is sponsored by the Institute for Information & Communication Technology Promotion (IITP), Korea.

This project is the 2nd phase of a project that was conducted at Kettering University from July 2019 to December 2019.

Research Initiation & Development Grant 2020

My research proposal, “Markerless 3D Human Motion Inference Framework from 2D Videos using Deep Learning,” has been selected for funding. This project is scheduled to start in May 2020 and end in December 2021.

https://umdearborn.edu/news/all-news/articles/update-office-research-april-2020

The project descriptions can be found at https://bimi.jrkwon.com/markerless-3d-human-motion-inference-framework-from-2d-videos-using-deep-learning/

Fig. 1. System overview. The proposed system has two sub-sections: 3D pose estimator from 2D video input (top) and gait feature extractor (bottom). (a) 2D input video, (b) 2D pose estimator, (c) extracted 2D joint points, (d) skeletal data of human pose, (e) skeletal model converter, (f) 2D joint points for 3D estimation, (g) 3D pose estimator, (h) 3D joint points, (i) feature extractor, (j) feature data, and (k) applications.

Ride On Car

A ride-on-car has arrived and assembled. We will add a drive, steer, and brake by wire to convert this ride-on-car to be a small scale autonomous vehicle.

Driving Simulator

A driving simulator hab been arrived and successfully installed. This will be used to study end to end learning and reinforcement learning for autonomous vehicles.