Tuesday, April 07, 2015

Everybody settle down.

I was at a conference a couple of months ago that really didn't have anything interesting to write about, BUT....the one of the interesting things I discovered was that Lidar had really come down in price. I've been watching this tech for years. Lidar gives robots the best depth perception of all the different types of sensors. It is also used in self driving cars. I used to think Lidar would win the sensor wars for robots, but now it seems clear robots will have lots of different sensors. And probably multiple Lidar sensors.

I bring this up because I used to think once the price of sensors came down, that would be the tipping point for robots. So, I was kinda excited to go to my conference today. Some Lidar sensors cost as little as 300 bucks now. You can buy old drum Lidar systems on ebay apparently!

After all this time, we must be on the verge of the tipping point by now. I read all these articles telling me that robots are coming to take our jobs. Which always makes me a little crazy. I go to a handful of these conferences a year. Robots are very complicated creatures. Mind bendingly hard. With many different modalities. Yet the way people write about them is just fanciful. It's taken them roughly 10 years to get to really good object avoidance.

Today after listening to all the speakers I left disappointed. Robots for the most part are still a solution in search of a problem. Outside of manufacturing and warehouse work. And worse, it seemed to me that the industry got a little over-hyped a handful of years back and promised too much, and now since they haven't been able to deliver on a huge level - it's harder to find venture capital. At least, that was the impression I got. Robotics in particular are massively expensive. A lot of those companies ate through a ton of cash and didn't deliver much that people would buy.

It's easy to see stuff on the internet and think we are closer than we are, but sadly we aren't as close as you think. No matter how much I want to be wrong. Are they going to take "some" jobs? Yeah - but not as many as you think. Not yet.

Now I wonder if self driving cars might be a missing piece of the puzzle. Because sensors are relatively super cheap. Cars are going to be learning how to recognize objects in the real world. Robots also need to learn about objects in the real world. And that is deeply complicated and expensive. Car makers have tons of money to crunch the insane amount of data that is going to take. This is my new tipping point.

Just as a side note I find it funny that people seem to be less scared of cars driving themselves than robots taking care of us. It's crazy how much energy this sector has to spend trying to make robots not scary.

I used to have a robot mowing my lawn with spinning blades of death, but that took several hours. It takes me 15 minutes. I also got  super self conscience of people just stopping in front of my house to watch it. Seriously.

No comments:

Post a Comment