Actually:
you are likely talking about DA Usikov who wrote the 1991 text "Three dimensional vision for Autonomous Vehicles: Algorithms, Problems, Perspectives."
I
Based on his body of work that I see, he is basically working on vision recognition problems, dealing (from what I can tell) with extracting 3d data from stereoscopic imaging systems. Which is all fine and dandy, but even if Mr. Usikov somehow knew what everyone else didn't, there are numerous ways to skin the cat of 'autonomous robotic vehicles' and many technologies are far more promising than Visual Recognition- namely Sonar, Radar, Lidar and IR.

I agree that "there are numerous ways to skin the cat of 'autonomous robotic vehicles'" but so far the Sonar, Radar, Lidar and IR systems have not performed very well in that function, so why not give the stereoscopic imaging systems a chance to prove what they can do.

Visual recognition as the 'holy grail' for robotics (or more precisely- for machines that interact with their environment) is often thought of as misguided. Its based on the observation: "well, thats how (most) animals work" Its like trying to build the first flying machine by trying to build an orinthopter and nothing else.

If the animals can do it, then why not imitate their function?

100 years later, we have hypersonic suborbital 'airplanes', but no orinthopters (and no interest in them).

Perhaps the reason the "orinthopters" are not in use today is because they did not function the way they were intended. Perhaps in 100 years the "Sonar, Radar, Lidar and IR" systems will be in a similar situation as far as machines that interact with their environment is concerned.

I doubt Mr. Usikov has some 'great breakthrough' not to be found in the public literature, and even if he did- its aort-of fantastic (in the fantasia sense) that he has a 'lock' on an 'algorithm'... Thats not how academia, patenting and innovation works in Computer Science. And even supposing he did, U of M and other places he worked would actuall own the 'lock'

Unless he is using the "BIONIC LOGIC DEVICE" patent number "CA 1013439" in the Canadian Patents Database.
The innovation was also patented in the United States, but I do not have the number for that patent right now.

Johannes

PS. If you want to know more about that patent then just ask me, I know all about it.