Japanese Robots: ASIMO's Nemesis Grows Stronger; Will Honda Respond? (VIDEO)
Honda bills ASIMO as the world's most advanced humanoid robot, and taken as a whole, it’s probably accurate to say. But an American robot is catching up, and unless Honda’s got some new tricks (ASIMO X?), ATLAS is going to shove ASIMO aside and take his cookies.
• • •
ASIMO is Awesome, but...
It’s safe to say a majority of humanity has had at least some sort of visual contact with one of Honda’s ASIMO robots. Introduced 13 years ago this month, the intentionally diminutive and non-threatening robots are beyond iconic. Although they primarily function as research platforms, the robots also make very decent corporate ambassadors and public speakers.
We saw the practical research side of things play out in rapid fashion with the development of Honda’s High-Access Survey Robot. The tracked robot has a top-mounted robotic arm derived directly from the the super-stable ankles used in ASIMO.
So, in addition to great PR and some practical offshoots, Honda's got this amazing robot capable of climbing and descending stairs, hopping on one leg, running, kicking a soccer ball, autonomous collision avoidance, it has pressure sensitive, articulated hands, the ability to listen to and parse three voices simultaneously and respond to human gestures; individually, these are very impressive engineering and computer science milestones; incorporating them all into a single machine, now that is a profound and heretofore unmatched technological achievement. Honda, now approaching 30 years of robotics and assistive mobility research, has done very well.
But it's not all adulation, and although we’ve stood in defense of ASIMO here in the past, the robot does deserve a measure of realistic criticism. See, when presenting ASIMO to the public, there are three important things that Honda glosses over, or doesn’t really mention at all:
- Many of ASIMO movements are remotely controlled by offstage engineers;
- ASIMO’s complex autonomous movements are only possible with placemarks and pre-set visual cues;
- Demonstrations are largely rehearsed and choreographed through familiar environments.
It would be pessimistic, disparaging, and an oversimplification to call ASIMO a super-duper high-tech puppet, but there is a measure of truth to that statement.
Dig, if you will, this analogy: think of ASIMO as the iPhone, an innovation released to great fanfare, hugely popular, best in the world for a lot of years and regularly updated with slight, subtle improvements. But, it’s not really surprising us anymore, and there’s been nothing really innovative or groundbreaking for a while. In a delicious bit of serendipitously aligned terminology, on the other side of the analogy, there’s Google’s Android operating system, their own Motorola, and Samsung, HTC, LG, et al; a small legion of manufacturers.
The analogy’s latter parties, they are the upstart humanoid robots competing in Track A of the DARPA Robotics Challenge (DRC): advanced, original machines, each with proprietary software, from NASA (x2), Carnegie Mellon University, Japan’s SCHAFT Robotics, Virginia Tech, and Korea’s KAIST with Drexel University; then there’s the biggest and arguably baddest kid on the block, Boston Dynamics' ATLAS, a standardized humanoid platform, one of which has been supplied to each successful DARPA Virtual Robotics Challenge qualifier (a preliminary stage of the larger DRC).
Okay, so what that means is that, not only are there 6 DRC Track A contestants threatening ASIMO’s #1 Robot in the World spot, there's also the threat posed by 6 competing teams creating software for the very intimidating and very capable ATLAS. In this new video, released just a few weeks back, we get a glimpse of what ASIMO's latest nightmares are made of:
Outclassing ASIMO Where it Really Counts
At 4 feet 3 inches and 119 lbs (130 cm/54 kg), ASIMO is the size of a child - with the frailty to match. ATLAS is 6 feet 2 inches and 330 lbs (187 cm/150 kg), and when shoved around, walking on a bed of large, irregular rocks, or, as the video above demonstrates, whilst standing on one leg, laterally impacted by a 44 lb/20 kg object, ATLAS just keeps on going. Poor ASIMO would surely topple and take damage.
For now, ATLAS lacks many of ASIMO’s refinements, but there are 7 world-class robotics teams working to change that - and they will catch up. And, ASIMO’s refinements are impressive, but a few years back when disaster hit northeastern Japan, and it seemed obvious that robots should help with response & recovery, ASIMO’s many refinements were useless; ironically, this actually spawned forth the aforementioned DARPA Robotics Challenge - a competition full of robots, including ATLAS, that all have the potential to relegate Honda’s currently #1 robot to the robot hall of fame; great for its time, but time to move on.
After Fukushima, in the Face of Huge Competition, Can Honda Respond?
Immediately after the 2011 Fukushima Daiichi nuclear disaster began, Honda received a flood of requests for assistance followed by no small amount of disappointment at their inability to offer anything but kind words. To many, it seemed obvious: if ASIMO is so advanced, well - hurry up - send one in already! Unfortunately, ASIMO is not advanced in the right ways, and sending one of the machines anywhere near the power plant would have accomplished little more than destroying an expensive robot.
Not only are ASIMO’s sensitive electronics vulnerable to high levels of radiation, but as follows from the criticisms outlined above, the robot isn’t autonomously agile enough to effectively navigate unknown environments. Moreover, even if ASIMO could get past debris and into a position to help, it lacks the strength to do things like clear debris, turn valves on/off, move levers, etc. (while unknown, it’s also likely that ASIMO cannot right itself after a fall, though no evidence for or against has been seen).
To be fair, there was no other robot - on earth - capable of assisting after the plant was flooded and compromised - and that's exactly why the DARPA Robotics Challenge came to be. If a sufficiently agile, sufficiently strong humanoid robot with radiation shielding could have entered the plant and turned a few valves, things would have turned out very differently. It's very likely that meltdown could have been prevented.
Yes, Honda will Step Up, and Akihabara News is Now Predicting: ASIMO X
Frustration and embarrassment at their powerlessness to help, in addition to the sting of 10 year-old American robots being the first able to enter the Fukushima disaster area, did spur Honda into action. Using technology developed for ASIMO’s ankles, by November of 2011 Honda had completed a new robotic arm with the dexterity, stability, and strength - even in actively unstable environments - to open/close standard pipe valves. Though impressive, the arm project was shelved for over a year until it found itself mounted on top of the High-Access Survey Robot (HASR), a machine jointly developed with government and private input that does exactly what it sounds like.
But also in November of 2011, Honda unveiled an upgraded ASIMO (the current touring model), and a new ‘Honda Robotics’ logo emblematic of efforts to bring all its projects under one robo-umbrella - the first such move in 26 years of robotics & mobility research.
We believe that Honda is working on something new, and they will respond. With timing likely aligned to the DARPA Robotics Challenge’s finale in December, 2014, what we're going to call "ASIMO X," or whatever Honda comes up with (probably a bad acronym), will go public.
• • •
VIA: Boston Dynamics YouTube; AkihabaraNews.com
Images: Honda; Boston Dynamics; AkihabaraNews.com