Robots are getting cleverer and more dexterous. Their time has almost come | Hacker News Reader

Robots are getting cleverer and more dexterous. Their time has almost come

economist.com

24 pointstheoneill12 comments

Hm, annoying - I live in Munich, and I would have loved to visit the robot fair, but I didn't know about it. What is a good way to not miss interesting conferences and fairs?

Friends of mine recently went to see the guided tour through the BMW factory in Munich and they said it was very impressive.

That was one I idea I had, wanted to work on, procrastinated it, and now I'm waiting for someone else to build the conference reminder.

I wonder if there is a conference on procrastination? ;-)

Probably later.

OMNI magazine in 1982 ran an article on the procastinator's club, which had been meaning to have meetings for the last couple of decades or so.

I've been thinking about joining. Still not sure yet, though.

(true story)

http://www.confsearch.org/confsearch/

"Almost come" is accurate. There is some major distance to be made in slew of robotics categories.

For instance, robots can still barely and accurately maneuver around using just vision. Slap on 3D like Lidar and it gets expensive both financially and computationally.

I'm sure I'm not the first/only one to have this idea, but to make the robot systems much cheaper, why not have an ecosystem of smart objects, and the robot itself much simpler? For example, for the food preparation industry, have all that food be in smart containers (self-weighing, able to dispence discrete amounts, chemical sensors to determine "freshness" of the food, if applicable). Is anyone aware of such systems, or perhaps working on them as part of a startup?

I've thought about such systems when hunting for startup ideas, but the snag that I kept coming across was: how do you convince people that they need all of these other 'smart objects' in addition to the people need to mesh the variety of smart objects.

On the other hand, most large manufacturers use robotic systems in the actual production of goods.

We couldn't even convert to the Metric system in America, I'm not saying its impossible but if you can't point to a clear benefit from it, why incur the cost?

I also think that right now the cost of robotics comes from the complexity of the software more so than the hardware. We have plenty of commodity hardware, and the actual robotics chassis/effector/sensor stuff is not all that expensive in perspective, but the software is a giant tangled mess of projects that do the same thing in different ways. (At least on the open source side). There have been moves lately to fix that, like the Gearbox project, but overall it is still a lot of people spending a lot of time reinventing the wheel.

The aggregate system of simpler machines will still need to get the task done. It's still a hard problem, no matter how you slice it. Systems probably also are less able to operate in human environments.

But one big trend in robotics is simpler robots that can solve part of the problem. We don't have butler robots, or systems of robots that are as capable as butlers, but we do have roombas & scoobas.

They only real-way to "Scale" robots, is to create a system of self-replicating robots. This way, they can organically interact with our (human) needs. Similarly to how one can raise a pig to eat, or, alternately, one can go hunt free range buffalo, one can custom-make a robot for one's particular need, or, instead, allow self-replicating "species" or robots to exist in well-defined areas. For example if one had a swarm of aquatic robots in say a harbor area, these could be remote-control programmed to fetch fish and bring those fish to a centralized net. The robots would replicate themselves, so no need to worry about fixing them or making more. With nano-tech who knows such might be possible. Effectively to maximize/economize robot use we need species "symbiosis" with robot strains. Everybody benefits. Without making robots become "organic"/self-replicating they remain to costly. However, carefully directing the evolution/creation of a symbiotic robotic ecosystem could benefit mankind. Robots are no threat, when handled properly.

Essentially two dangers lurk with robots: one they are economically prohibitive. But flash forward a hundred years and you have a different problem: controlling them. Directing the evolution of robots could solve both problems. In the short term make them more cost effective vis-a-vis population growth, in the long term create symbiotic interdependence with the robot species, by means of orchestrating and directing their initial evolution. It is like a flash flood. You can't stop it, but you can control where the waters go. Similarly we cannot stop the "singularity" the rise of the robots constitute, but we can control it. (Me, my ultimate dream is a robot drag queen, yeah, when I make my millions I'm gonna build me one of those and knock 'em dead at karoake night, lol!)

Reply on news.ycombinator.com