cold_water
8年前
. . Corporate HQ is his mother-in-law's house, is a false rumor.
The Spencers do not live in their mother-in-laws home.
That is why:
7.
Neil Wallace, brother of Reed Wallace, has spread false information about the Plaintiffs' company on the internet, to the courts, its attorneys and in numerous correspondences at times using his parents and family to assist in these efforts directly and indirectly. His public disparagement of the company has put false rumors, and therefore doubts, in the minds of the Plaintiffs' business associates, potential customers, attorneys, and stock investors. http://finance.yahoo.com/news/geckosystems-praises-scotus-decision-naked-040000765.html
XenaLives
8年前
I think GOSY's biggest problem is one of perception - the tech works but so many academics and companies have failed. I once found a Master's dissertation on how depth cameras generated a point cloud that was too dense to be used for navigation purposes.
The following is an OCR and may contain some errors, but it makes my point. Personal service robots such as the CareBot are gated by navigation issues. (See bold text.) Baxter is gated by navigation issues with the arms. This is probably why Market and Markets continues to list GOSY's tech in their reports. No one else can do it.
Towards Coexistence of Human and Robot: How Ubiquitous Computing Can Contribute?
Jingyuan Chengl, Xiaoping Chen’, and Paul Lukowiczl
' German Research Center for Arti?cial Intelligence (DFKI),
'Ii'ippstadter Strae 122, D-67663, Kaiseislautern, Germany
2 Multi-Agent Systems Lab, University of Science and Technology of China,
.]inzhai Road 96, Hefei, 230026, China
Abstract. After the ISO 10218-1 / 2 in 2011, safety factors for industry
robot are standardized. As robotics expands its area from industry fur-
ther into service, educational, healthcare and etc., both human and robot
are exposed to a space with more openness and less certainty. Because
there is no common safety speci?cation, we raise in this paper our own
hypotheses on the safety requirements in dense human-robot co-existing
scenarios and focus more on demonstrating the possibilities provided by
the research ?eld named Ubiquitous Computing.
1 Introduction
The number of robots over the world keeps on growing. According to the World
Robotics studies [1], 159,346 units of industry robots were sold in 2013, 16,067
professional service robots and about 3,000,000 personal and domestic use robots
were sold in 2012. As robots’ population grows, the physical even emotional con-
tacts between robots and human are also growing, freeing human from certain
labor work and meanwhile bringing potential risks. As in early ages robots
were implemented mainly in industry, safety speci?cations have been (level-
oped mainly for industrial robots, e.g. ISO 10218-1 /222011 sets the rule on both
robot itself, the robot system and integration [2] [B]; in US the under revision
ANSI/RIA R.l-5.06-2012 re-opens the allowance of human and robot working in
a loop. In industry, the general trend shifts from strict isolation of robot from
human to detailed speci?cations on reducing hazards. Out side of industry, how-
ever, robots are already in close contact with human, especially in the case of
service robot. Since there is no ?xed global speci?cation till now, researchers fol-
low their own ideas on whether and how to separate robot from public audience
at Expo’s and demonstrations. Domestic-use robot providers take care of human
by hiding rigid components inside, reducing the robot’s weight and speed, and
implementing obstacle detecting sensors.
Ubiquitous computing as a fast developing ?eld focuses on pushing the one
central powerful computer to multiple little computing units into the environ-
ment and onto human beings. This indicates tiny sensing al1d processing units,
and naturally, the application ?eld of environment and human activity monitor-
ing, which are also important factors a robot might need, when comes to dense
contact with human beings.
Hazards in Dense Human-Robot Co-existing Scenarios
Vasic and etc. gave a detailed survey on safety issues in human-robot interac-
tions Starting from industry, the danger comes when human gets trapped
between robot and an object (e.g. a wall) or when human cronies into collision
with a robot A detailed list of signi?cant hazards can be found in ISO 10218-1
as annex, including: Mechanical, electrical, thermal, noise, vibration, radiation,
material /substance, ergonomic, the hazards associated with environment and
combined hazards. The hazard should be analyzed and minimized from techni-
cal points of view, however, in the real applications, there are still unexpected
errors and failures which can not be exactly predicted:
— mechanics failure: aging of motors, connectors;
— electronics failure: aging of components and isolation material, out of power
half the way of operation;
— program failure: program hugs, untested scenarios;
— operational error: untrained engineers, operators, and users;
Besides regular maintenance, the above listed hazards are minimized in in-
dustry applications by:
— strictly pre-de?ned environment and space (robot cell);
— strictly pre-de?ned operation routine;
— authorization of properly trained operators, maintenance workers and pro-
gramers;
— speed limitation when human is present;
— protective stop function and an independent emergency stop function.
While robot go out to factory and into family or other social areas, the above
conventional rules become invalid. The situation is similar to that of computer
going from military use to civil and then personal use. The difference is however
the actuation, the capability of active physical movement brings more potential
hazards. Moreover, the robot enters an open environment where changes may
happen anytime and anyhow, the users are most often non-professional and
unexperienced people. Animals (e.g. pets) might come into close contact with
the robot, which might even bring damage to robot. (e.g. a child might see a
home service robot and pour water onto it just out of curiosity.) Due to these
obstacles, the most sold service robot now is still household robots, which are
small in size, carry out comparatively simple and ?xed tasks.
link.springer.com/chapter/10.1007/978-3-319-16841-8_39
XenaLives
8年前
November 2015 - Grad student FAILS
Towards Vision-Based Deep Reinforcement Learning for Robotic Motion Control
Fangyi Zhang, Jürgen Leitner, Michael Milford, Ben Upcroft, Peter Corke
(Submitted on 12 Nov 2015 (v1), last revised 13 Nov 2015 (this version, v2))
This paper introduces a machine learning based system for controlling a robotic manipulator with visual perception only. The capability to autonomously learn robot controllers solely from raw-pixel images and without any prior knowledge of configuration is shown for the first time. We build upon the success of recent deep reinforcement learning and develop a system for learning target reaching with a three-joint robot manipulator using external visual observation. A Deep Q Network (DQN) was demonstrated to perform target reaching after training in simulation. Transferring the network to real hardware and real observation in a naive approach failed, but experiments show that the network works when replacing camera images with synthetic images.
Comments: 8 pages, to appear in the proceedings of Australasian Conference on Robotics and Automation (ACRA) 2015
Subjects: Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
Cite as: arXiv:1511.03791 [cs.LG]
(or arXiv:1511.03791v2 [cs.LG] for this version)
Submission history
From: Fangyi Zhang [view email]
[v1] Thu, 12 Nov 2015 06:19:59 GMT (4489kb,D)
[v2] Fri, 13 Nov 2015 05:41:08 GMT (4489kb,D)
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
http://arxiv.org/abs/1511.03791
SafePath navigation could be applied to the arms of this unit and they would not fail....
XenaLives
8年前
Yeah, a grade school project that does what no other navigation system can do.
SafePath can't run over little kids.
http://www.dailymail.co.uk/news/article-3686411/He-crying-like-crazy-300lb-mall-security-robot-used-catch-shoplifters-runs-one-year-old-boy.html
http://gizmodo.com/mall-suspends-security-robot-after-it-clobbered-a-toddl-1783662510
Interesting quote from above:
In response, the Stanford Shopping Center in Palo Alto, California, is investigating the incident, and it has docked its robots until further notice. The company that developed the robot, Knightscope, called the incident a “freakish accident” and has issued a formal apology to the family.
“Our primary mission is to serve the public’s overall safety, and we take any circumstance that would compromise that mission very seriously,” said William Santana Li, Knightscope’s chief executive.
The company claims that the machine, a K5 unit equipped with nearly 30 sensors, should have registered a vibration if and when it ran over Harwin’s foot. The incident prompted Knightscope to file a field incident report—the first such report after 25,000 miles of total travel made by its robotic fleet. The company’s account of the incident varies a bit from that of the mother’s. Knightscope claims that Harwin ran backward “directly” into the machine, prompting the robot to stop, and that’s when the child fell.
Regardless of what happened, there’s no disputing that a child got hurt, even if it was a minor injury. Malls may find it cute and expedient to have robotic security guards roaming the corridors, but they clearly need to take account of these mechanical lumbering beasts which weigh 300 pounds and stand five feet tall.
A 300 pound robot has to run over someone's foot before they even expect it to do something? It's like a bull in a china shop - do you think the owner of this shop would let that robot roam free like this inside the shop? Yes - an early SafePath robot avoiding things before it runs them over, what a novel idea.
http://geckosystems.com/timeline/?year=2003&video=CoinShop-Display_avoid.mpg
SafePath can't run away and become so dangerous it has to be "put down".
http://www.bbc.com/news/blogs-news-from-elsewhere-36547139
Tesla does its Beta testing on customers and they get killed - oops missed seeing that semi because it was white. GOSY did its Alpha testing on a beloved member of the family for months to make sure it was safe.
Tesla puts the responsibility for Beta testing a flawed system on its customers.
The Autopilot feature was engaged at the time, Tesla has said, but neither the automatic braking system nor Mr. Brown applied the brakes before the car hit the trailer at 65 miles an hour.
Despite that acknowledgment by the company, as the federal agency pushes for answers about the accident and whether the Autopilot system failed to work properly, Tesla officials continue to say that the technology is safe. They also say they have no plans to disable the feature, which is installed in about 70,000 Tesla cars on the road. Instead, they indicate that drivers may be to blame for misusing Autopilot.
In an interview, a Tesla executive said the company believed that the system was safe as designed, but that consumers needed to realize that misusing Autopilot “could mean the difference between life and death.”
http://www.nytimes.com/2016/07/13/business/tesla-autopilot-fatal-crash-investigation.html?_r=0