I recently participated in a two week class offered by MIT on autonomous cars (https://selfdrivingcars.mit.edu). This class consisted of lectures by instructor Lex Friedman as well as sessions by people from industry working in this topic. One thing was clear — there is renaissance in Automotive industry now. This is an industry that until few years ago was considered an old and mature industry. I have interacted with numerous automotive companies in the past As part of job. It was all about process improvement, cost reduction etc. While all those things are still important, this is different now. It is all about innovation and dramatic changes this technology is going to bring. It is quite interesting that something that something as invisible as software came to make this old hardware focused industry hot again. In fact, as I spend more time to understand the impact of autonomous vehicles the prognosis is nothing but mind boggling.
One of the key developments that facilitated it has been how deep learning improved machine vision by leaps and bounds. Object recognition through cameras is relatively easy to do. Yes, there are ways to spoof them and inaccuracies are quite common, but the trend is clear. The breakthrough in AI enabled made autonomous driving good enough for street driving, which was considered a hard problem.
As I was pondering over this, I realized that the impact this extends to a number of industries adjacent to cars. After all automobile was “the machine that changed the world” and autonomous driving could change all of that again. I would like to spend some time in the next posts to explain my thoughts on this topic.
To be continued….
Source: Deep Learning on Medium