His name is Murata Boy, and it's hard to believe that he can ride a bicycle better than any person or any other robot. It is equally hard to believe that this robo-bicyclist robot has been out there for five years.
The Murata Boy robot only came to existence because the Japanese company Murata Manufacturing wanted to show off some of their various sensors technologies, hence the name. The robot made his first debut at CEATEC (Combined Exhibition of Advanced Technologies) Japan 2005 and people immediately fell in love with this little adept robot who can ride a bicycle like no robots before.
The Murata Boy is only 2-foot tall, which means he can only ride the tiny bicycle custom-built for him. The robot is equipped with four types of sensors: two gyro sensors used to detect angular velocity and inclination; an ultrasonic sensor to detect obstacles; and a shock sensor to detect rough surfaces (you can see more detailed specifications at the project web page here). With wireless network connection to a PC, the robot can follow remote commands such as forward, stop, and reverse. It can also follow pre-programmed paths. The robot can reach maximum speed of 30 inches per second and can even balance while at a complete stop. The robot's sensors work so well that the robot can even ride on a very thin balance beam as show in the video below (toward the end of the video).
Ever since its first debut in 2005, the robot has become very popular at technology exhibitions. And throughout the years, it also had to take on additional challenges with the help of the team of developers. The newly acquired skills include ones such as riding on an S-shaped beam and going uphill on a quite steep slope. Such tasks might sound easy, but requires precise sensor measurements and very accurate motor control. The end result, a very amazed crowd whenever the robot goes. (Change the video resolution to 480 for the two videos below for better viewing experience.)
Murata Boy riding and balancing on an S-Shaped beam
Murata Boy riding uphill
Although the robot is not capable of complex autonomous behaviors (not something Murata Manufacturing wants to show off), it certainly gives a good demonstration of what the sensors are capable of. Once we can develop advanced intelligence to combine with the wonderful engineering feats, who knows, maybe once day we'll see robots throwing morning newspapers at your doors. That is, of course, after the robot is equipped with dog-repelling (or evading) mechanisms. But anyhow, respect to the engineering team behind the Murata Boy!
The team of engineers at Murata Manufacturing who developed the Murata Boy robot
It's worth mentioning that some people opted to turn real bikes directly into robots. For example, the first video below shows the effect of some Japanese inventors working on a noodle-delivering unmanned bike. And the second video shows a Vietnamese attempt. Just imagine a fleet of unmanned bikes on the streets of Beijing, China, taking people to their work! What an amazing scene that would be! Here in the US, however, an unmanned bike is likely to end up an exercise equipment so the robotic bike will do the exercise so the lazy fat human won't have to do it. LOL!
Video of the Day:
Vuvuzela Concerto in Bb. If you click the video to open it in YouTube, you can also click the soccer button in the lower right corner to make it a real concerto!! And here's the music score that goes with it.
Linghu Chong was dumbfounded. He had been concentrating so much on the energy cultivation and dispersion that he didn’t even realize that three days had passed and that it was the day he told Mr. Black-White to come back. He didn’t even notice Mr. Black-White’s arrival. Luckily his voice was so coarse now that Mr. Black-White didn’t find anything unusual. He quickly let out several hollow laughs as an acknowledgement.
“The revered mister seems to be in a pleasant mood today. Why not make it the day the revered mister receives a new apprentice?” Mr. Black-White suggested.
Linghu Chong thought to himself, “If I agree to take him in as an apprentice and teach him these energy cultivation techniques, as soon as he opens the door and comes in, he’ll notice that there’s only Feng Er-Zhong, but no revered mister Ren. For sure he’ll have a fall out immediately. Besides, even if it were the real revered mister Ren, once Mr. Black-White learns those techniques, he’ll most likely manage to murder revered mister Ren, like poisoning the food or something. Right! It would be so easy for Black-White to murder me with poison. Why would he want to let me escape after he learns the secret formulas? No wonder revered mister Ren wouldn’t ever agree to teach him the secret art in the last twelve years.”
“Once revered mister teaches me the magical art, the humble apprentice will show his respect by serving the revered mister great wine and tasty roast chicken,” not hearing a reply, Mr. Black-White immediately proposed.
Linghu Chong had been imprisoned for many days and was only given plain vegetables and tofu each day. As soon as he heard the words “great wine and tasty roast chicken,” his mouth drooled with greed. This was too tempting!
“Fine. Go get me some great wine and tasty chicken first. If they get me into a good mood, I might teach you something.”
“Sure! Sure! I’ll bring back some great wine and tasty chicken. But I am afraid it won’t be today. If opportunity allows, I’ll bring them in tomorrow,” Mr. Black-White replied hurriedly.
“Why not today?” Linghu Chong asked.
“To get in here, I have to pass through my eldest sworn-brother’s bedroom. Only when he is away, I could…could….” Mr. Black-White explained, to which Linghu Chong let out a snort without saying another word.
Afraid that Mr. Huang-Zhong might return to his bedroom, Mr. Black-White dared not to stay any longer. He quickly said his goodbye and then went away.
“How can I trick Black-White into the prison cell and whack him?” Linghu Chong thought to himself. “This man is so crafty; it would be impossible to swindle him. Moreover, without a way to break these iron shackles and chains around my wrists and ankles, even if I succeed in killing Black-White, I would still be stuck here.”
With that thought, his right hand naturally reached for his left wrist and then pulled. This was only a spontaneous move, and he certainly didn’t expect the iron shackle to break open, but to his great surprise, the iron shackle did actually open. He gave it a few more hard pulls and even managed to free his left wrist from the shackle.
This was definitely not something Linghu Chong could foresee, and his heart thumped quickly from the mixed feeling of shock and delight. He examined the iron shackle with his finger and then found a slit in the middle of the ring. If he had not dispersed the inner energy streams inside him, any hard movement would have made him faint. Even with the slit in the middle of the shackle ring, he would not have been able to pull it open. Since he had been working on dispersing inner energy streams for over two days, the energy streams the Peach Valley Six Fairies and Monk No-Commandment had forcefully injected into his body have been diffused into his Ren Channel, which in turn spontaneously originated strong inner force.
He felt the iron shackle on this right wrist. Sure enough, there was also a thin slit. He probably had felt the thin slit many times before, but never in his mind associated it with a crack in the shackle ring. He pulled hard with his left hand and freed his right hand from the shackle ring. Next he felt the two shackle rings around his ankles and also found thin slits in both. After some more hard pulling, he found himself breathless, covered in sweat, but completely free from all the shackles.
Once rid of the shackles, the iron chain no longer restrained him in anyway. He was truly puzzled: “Why is there a slit on each of the shackle ring? How could this kind of shackles have fettered anyone?”
The next day, when the deaf and mute old man delivered his food, Linghu Chong was able to actually look at the shackles closely. Next to each of the slits was a fine-grained saw pattern. Obviously someone had cut through the four shackles rings with a very thin steel wire saw. The slits shined under the dim light and there were no rust on the fractures. This meant that the cuts must have been made only recently.
“But why were these shackle rings closed again around my limbs?” Linghu Chong asked himself. “Someone must have tried to free me secretly. This underground dungeon is so well hidden; no one from the outside would be able to break in. Therefore, the secret rescuer must be someone inside the Plum Manor. Maybe he disliked the plot against me. That’s why he secretly cut open my shackles with a steel wire saw while I lay unconsciously in the prison cell. He probably didn’t want to openly declare himself an enemy of the Plum Manor, and he is still looking for an opportunity to help me escape.” At the thought, his spirit was greatly lifted.
“The entrance to the underground tunnel is concealed underneath Mr. Huang-Zhong’s bed in his bedroom. If it were Mr. Huang-Zhong who had wanted to rescue me, he could have done it anytime. He didn’t have to delay. Mr. Black-White obviously can’t be the one. Out of Mr. Bald-Brush and Mr. Paint, Mr. Paint is a confidant in the art of wine and had good fellowship with me. I bet it must be Mr. Paint.”
Then he remembered that Mr. Black-White would return the next day and thought about how he should cope with him. “I’ll just gloss things over with him. Why can’t I fool him with fake Kung Fu in exchange for some wine and meat?”
Then he thought, “Mr. Paint could come and rescue me any day now. I’ve gotta hurry and memorize all these formulas and techniques on the steel plate.”
He traced the carved characters and then recited each word carefully. When he studied these characters before, he didn’t pay very close attention, so memorizing everything with one hundred percent accuracy was no easy task. The handwritings on the steel plate were hasty and careless. With his limited education, he couldn’t recognize some of the poorly written characters. So he forced himself to memorize the strokes of the character and then randomly picked characters with similar strokes as substitutes. He knew that formulas for first-class martial arts are serious matters. Even one incorrect character could mean difference between life and death, success and failure. Any slight misunderstanding could very well lead to fire-deviation. Once he got out of the prison, who knows when he’ll be able to come back and reference the steel plate again? He must make sure he memorized everything with absolutely no errors.
So he read them again and again, losing count of how many times he had recited them, until he felt that he could even recite backward from the end to the beginning. Only by then, he felt at ease and allowed himself to fall asleep.
That night, in his dream, Linghu Chong actually saw Mr. Paint coming forth to open the prison cell door and let him out. The excitement woke him and then he realized it was only a fond dream. He didn’t let that depress him.
“He hasn’t come to rescue me today only because he doesn’t have the right opportunity. He’ll certainly come to my rescue before long,” he thought to himself.
He figured that although the formulas and techniques carved on the steel plate were very beneficial to him, they could be extremely harmful to others. If someone else gets locked up in this dark prison cell again in the future, that someone must have been a good person, and he wouldn’t let that person fall for Ren Woxing’s scheme. With that in mind, he traced the handwritings and recited the entire thing for another ten time or so, and then scraped off over a dozen characters in the passage from the steel plate using the iron shackles he freed himself from.
Mr. Black-White didn’t show up that day, but Linghu Chong didn’t care much about it. Instead, he continued cultivating his internal energy streams according to the formulas and methods on the steel plate. For the next several days, Mr. Black-White never came. Linghu Chong could tell that he had made great improvement in his training. Over sixty percent of the heterogeneous energy steams left inside him by the Peach Valley Six Fairies and Monk No-Commandment had been successfully dispersed out from his Dan-Tian and into his various Ren and Du Channels. He was sure that all of them could be dispersed out if he persevered.
Each day, he would recite the formulas scores of times and then scrape off a dozen or so characters from the steel plate. He noticed a gradual increase in his strength, and it became easier and easier to scrape the steel plate with the iron shackles. Another month went by in this manner. Although he lived deep underground, he could still sense a gradual decrease of the sweltering summer temperature.
“This must have been the work of divine providence. If I had been imprisoned here during winter, I would never have found the handwritings on the steel plate. Who knows? Perhaps before summer hits, Mr. Paint would have already had me rescued.”
At that moment, he suddenly heard the sound of Mr. Black-White’s footsteps from the outside tunnel.
Now support the translator Lanny by following my blog and leaving comments! :)
Video of the Day:
This is another great scene from the movie Fist of Legend by Jet Li. The moves he used here are typical Qin Na (grappling) techniques.
This paper was written by Hanjoon Kim and Sanggoo Lee from Seoul National University presented at the ninth international conference on Information and knowledge management, 2000.
Traditional document clustering methods are unable to correctly discern structural details hidden within the document corpus because they inherently strongly depend on the document themselves and their similarity to each other. To solve this problem, this paper proposes a clustering method that incorporates relevance feedback into a traditional complete-linkage clustering algorithm in order to find more coherent clusters.
The main contribution of this paper is the method of combining complete-linkage hierarchical agglomerative clustering (HAC) algorithm using the VSM model with user relevance feedback using the fuzzy information retrieval. The HAC with VSM allows the forming of pre-clusters using a “fixed-diameter”. Then documents in small clusters (less than η) form a “training document set”, which is used for user relevance feedback where user can answer “yes/no” questions to indicate relevance of documents returned using fuzzy information retrieval. This way, positive and negative bundles can be created and used to help reassign pre-clusters. Note that all these techniques mentioned above are not original ideas from the authors. The paper simply proposed a way to combine the strengths of these techniques in order to improve performance of the clustering.
Toward the end of the paper, the authors included two subsections that discuss the sensitivity of the degree of supervision parameter and the pre-cluster size parameter (“fixed-diameter”). This is a good attempt to justify the “magic numbers” used in the experiments. However, there is no discussion or analysis of the parameter η used in the supervising phase at all. Also, the paper describes that, “the threshold diameter value for pre-clusters is experimentally adjusted to provide good quality clustering.” This makes the reader wonder whether the experimental results presented are truthful and question how well the proposed method would work in real world applications.
Clustering techniques are used to group together things that have common themes. However, frequently, there are many different ways to group things based on different common themes and the distance function used decides what common theme will be used for the clustering. For example, a bunch of people can be grouped by one of many themes (e.g. age, gender, race, wealth, and hair style). Similarly, a document collection can also be clustered by themes such as style, topic, language, or subject. The paper simply stated the two alternative ways to define the diameter of a cluster but didn’t analyze how such selections might affect the results of the document clustering. The paper also did not clarify which definition of the diameter is used in the experiments.
The authors mentioned in several places that the proposed method helps in discerning structural details hidden within documents. However, throughout the paper, there is no discussion about how structural details are extracted from documents. Arguably the users might be able to identify structural details through user relevance feedback. However, this really has nothing to do with the proposed algorithm. The users might not even be looking at the structural detail of these documents at all.
A core part of the proposed method relies on human to provide domain-specific knowledge for the method to work, yet the paper is missing detailed description of how the experiment was designed to incorporate human knowledge. For example, we don’t know how many human analysts are involved. Could the human analysts have different opinions? If so, how is that handled? Are these human analysts really domain experts who possess domain-specific knowledge about the subject? The paper is also missing detailed description about how the queries are reformulated, which is an integral part of the user relevance feedback method.
In the supervising phrase, the user is asked to answer “yes/no” questions to mark documents as relevant to non-relevant to the query document. Firstly, relevance is a subjective question. It is possible that document marked as relevant by one user might end up being marked as non-relevant by another user. Secondly, a binary answer completely ignores the level of relevance (ranking). This results in that documents border lining non-relevance could still be marked as relevant and have the same weight as the most relevant document. This leads to inaccuracy and misrepresentation. A non-binary weighting probably would work better.
When building the test document set, the authors intentionally picked those that had a single topic, claiming that this would avoid the ambiguity of documents with multiple topics. Is this really for technical convenience so the experiment results would look better? Does this mean the proposed method would not work well at all for documents with multiple topics (which is probably more like the real world scenarios)?
The authors used F1-measure as the performance metric and used the complete-linkage clustering method as a baseline. The baseline selected is reasonable in order to show how much benefit can be achieved by adding the user relevance feedback component. However, using this baseline alone is not very convincing because we already know the complete-linkage clustering method doesn’t work well, and we also know that adding user relevance feedback will certainly help. It would be much more convincing if the paper compared the performance of the proposed method with another common document clustering technique such as K-Means and Bisecting K-Means. Using only the F1-measure as the performance metric is also weak. Including other quantitative metrics together with some qualitative analyses would certainly help.
In summary, this paper proposes a novice approach that combines the strengths of several IR techniques. However, it also has plenty of room for improvement.
Picture of the Day:
The biggest failure of the 2010 World Cup -- Domenech! We still wonder, "Why not Henry and why not Malouda?"
This mini-robot has got to be the cutest robot I'd ever run across, and her name is HINA, made by Mujaki Clockwork (a secretive Ninja name of some sort, and here's the link to his blog in Japanese, which I can't understand). As soon as I laid my eyes on her, I couldn't help but think: "I so must get my daughter one of these!!"
HINA is not one of those cheap robot toys you can just buy at Wal-Mart, she is a real robot that has servos with many degrees of freedom and can run complex scripts. Besides being cute always, she can also be quite a little helper sometime and can even provide maid services such as making you a cup of morning coffee!
The above video is obviously staged, but it does shed some lights about what the robot is capable of. Hina is made using Kondo's KHR-2HV, a Japanese robotic kit that could cost $1,299.99 in the US, but customized by Mujaki to give her a very adorable look. She is 36 cm tall and weighs 1.1 kg. With a total of 21 DOFs (Degree of Freedoms) -- legs 6x2, arm 4x2, neck 1 -- she is capable of walking, lifting things up and moving them around. HINA is also equipped with 2 gyroscopes and acceleration sensor, which makes her a real robot, just not a very sensitive one. But with some further customization, it wouldn't be difficult to add a mini-webcam, some touch sensors, and two microphones to give her more capabilities to learn about the world around her. Laser and sonar would probably be out of the question for her tiny slim figure, which means you might need racks of server to process data collected from these sensor in order to produce some intelligent autonomous behaviors.
But still, isn't she the cutest robot in the planet?! And besides providing maid services, she can also entertain you with some cute break dancing moves! So what are you waiting for?
Watching World Cup games is another great way to put your 1.5 year old son to sleep (it might have something to do with the constant bee-like buzzing sounds from the vuvuzelas) .
This paper was written by Comaniciu and Meer of Rutgers University presented at International Conference on Computer Vision 99.
In order to avoid estimation of the probability density function, mean shift is applied as a nonparametric estimator of density gradient. In image segmentation as well as smoothing process, it’s difficult to keep details while merge similar parts solely by spatial domain or range domain solutions. Usually users are required to set many constraint parameters for these tasks.
This paper proves the convergence of mean shift on lattices, and employs mean shift in the joint, spatial-range domain of gray level and color images for discontinuity preserving filtering and image segmentation. In processing in Spatial-Range Domain, each data point becomes associated to a point of convergence which represents the local mode of the density in the d-dimensional space. The output of the mean shift filter for an image pixel is defined as the range of information carried by the point of convergence. For the segmentation task, the convergence points sufficiently close in the joint domain are fused to obtain the homogeneous regions in the image. And for smoothing task, these points are merged together.
Mean-shift applied to several videos by someone somewhere (not me).
Based on proof of the convergences and applications in image smoothing and segmentation, mean shift can merge similar regions while keeping high details at the same time. Also this method reduces parameters needed for these tasks. What is more, the mean shift has a more powerful adaptation to the local structure of the data, and can run until convergence without stopping criterion.
Picture of the Day:
The differences between the World Cup and the Ph.D., (credit: http://www.phdcomics.com).
The Murata Boy is only 2-foot tall, which means he can only ride the tiny bicycle custom-built for him. The robot is equipped with four types of sensors: two gyro sensors used to detect angular velocity and inclination; an ultrasonic sensor to detect obstacles; and a shock sensor to detect rough surfaces (you can see more detailed specifications at the project web page here). With wireless network connection to a PC, the robot can follow remote commands such as forward, stop, and reverse. It can also follow pre-programmed paths. The robot can reach maximum speed of 30 inches per second and can even balance while at a complete stop. The robot's sensors work so well that the robot can even ride on a very thin balance beam as show in the video below (toward the end of the video).
