Path planning & AI

It is the final step, path planning, that has proved difficult. This is
the ‘doing’ step (as opposed to the previous two ‘learning’ steps) and is
also in some ways the most critical. If companies do not perfect this step
before putting cars on the road, accidents could sway public opinion and
force politicians to impose regulation on the growing sector.

Path planning involves using the data from the first two steps and then
integrating it with the learning from the artifcial intelligence. This then
integrates with the vehicle’s position coordinates and endpoint goal.
Altogether, the system navigates around potential obstacles and predicts
how other moving vehicles will react and what that means for the amount
of free space available to drive the car.

The other catalyst for a 2018 surge in autonomous driving are
new chip and systems. Intel and MobileEye have developed their own
system while Nvidia will release its Xavier chip this year. This will
integrate both a GPU unit that is extremely useful for deep learning, and a
regular processing unit that can be more suited for decision making. As a
result, more carmakers will take their autonomous cars out of the testing
phase and onto the road. The implication for companies that make these
types of chips and systems is signifcant.

In fact, the total addressable
market could double each year for the next four years and by 2021, these
systems will likely be a billion dollar industry.

AI is likely to be the next frontier for prospective investors,
watch this sector.