Category Robots Podcast

By Rob Matheson

A novel model developed by MIT and Microsoft researchers identifies instances in which autonomous systems have “learned” from training examples that don’t match what’s actually happening in the real world. Engineers could use this model to improve the safety of artificial intelligence systems, such as driverless vehicles and autonomous robots.

The AI systems powering driverless cars, for example, are trained extensively in virtual simulations to prepare the vehicle for nearly every event on the road. But sometimes the car makes an unexpected error in the real world because an event occurs that should, but doesn’t, alter the car’s behavior.

Consider a driverless car that wasn’t trained, and more importantly doesn’t have the sensors necessary, to differentiate between distinctly different scenarios, such as large, white cars and ambulances with red, flashing lights on the road. If the car is cruising down the highway and an ambulance flicks on its sirens, the car may not know to slow down and pull over, because it does not perceive the ambulance as different from a big white car.

In a pair of papers — presented at last year’s Autonomous Agents and Multiagent Systems conference and the upcoming Association for the Advancement of Artificial Intelligence conference — the researchers describe a model that uses human input to uncover these training “blind spots.”

As with traditional approaches, the researchers put an AI system through simulation training. But then, a human closely monitors the system’s actions as it acts in the real world, providing feedback when the system made, or was about to make, any mistakes. The researchers then combine the training data with the human feedback data, and use machine-learning techniques to produce a model that pinpoints situations where the system most likely needs more information about how to act correctly.

The researchers validated their method using video games, with a simulated human correcting the learned path of an on-screen character. But the next step is to incorporate the model with traditional training and testing approaches for autonomous cars and robots with human feedback.

“The model helps autonomous systems better know what they don’t know,” says first author Ramya Ramakrishnan, a graduate student in the Computer Science and Artificial Intelligence Laboratory. “Many times, when these systems are deployed, their trained simulations don’t match the real-world setting [and] they could make mistakes, such as getting into accidents. The idea is to use humans to bridge that gap between simulation and the real world, in a safe way, so we can reduce some of those errors.”

Co-authors on both papers are: Julie Shah, an associate professor in the Department of Aeronautics and Astronautics and head of the CSAIL’s Interactive Robotics Group; and Ece Kamar, Debadeepta Dey, and Eric Horvitz, all from Microsoft Research. Besmira Nushi is an additional co-author on the upcoming paper.

Taking feedback

Some traditional training methods do provide human feedback during real-world test runs, but only to update the system’s actions. These approaches don’t identify blind spots, which could be useful for safer execution in the real world.

The researchers’ approach first puts an AI system through simulation training, where it will produce a “policy” that essentially maps every situation to the best action it can take in the simulations. Then, the system will be deployed in the real-world, where humans provide error signals in regions where the system’s actions are unacceptable.

Humans can provide data in multiple ways, such as through “demonstrations” and “corrections.” In demonstrations, the human acts in the real world, while the system observes and compares the human’s actions to what it would have done in that situation. For driverless cars, for instance, a human would manually control the car while the system produces a signal if its planned behavior deviates from the human’s behavior. Matches and mismatches with the human’s actions provide noisy indications of where the system might be acting acceptably or unacceptably.

Alternatively, the human can provide corrections, with the human monitoring the system as it acts in the real world. A human could sit in the driver’s seat while the autonomous car drives itself along its planned route. If the car’s actions are correct, the human does nothing. If the car’s actions are incorrect, however, the human may take the wheel, which sends a signal that the system was not acting unacceptably in that specific situation.

Once the feedback data from the human is compiled, the system essentially has a list of situations and, for each situation, multiple labels saying its actions were acceptable or unacceptable. A single situation can receive many different signals, because the system perceives many situations as identical. For example, an autonomous car may have cruised alongside a large car many times without slowing down and pulling over. But, in only one instance, an ambulance, which appears exactly the same to the system, cruises by. The autonomous car doesn’t pull over and receives a feedback signal that the system took an unacceptable action.

“At that point, the system has been given multiple contradictory signals from a human: some with a large car beside it, and it was doing fine, and one where there was an ambulance in the same exact location, but that wasn’t fine. The system makes a little note that it did something wrong, but it doesn’t know why,” Ramakrishnan says. “Because the agent is getting all these contradictory signals, the next step is compiling the information to ask, ‘How likely am I to make a mistake in this situation where I received these mixed signals?’”

Intelligent aggregation

The end goal is to have these ambiguous situations labeled as blind spots. But that goes beyond simply tallying the acceptable and unacceptable actions for each situation. If the system performed correct actions nine times out of 10 in the ambulance situation, for instance, a simple majority vote would label that situation as safe.

“But because unacceptable actions are far rarer than acceptable actions, the system will eventually learn to predict all situations as safe, which can be extremely dangerous,” Ramakrishnan says.

To that end, the researchers used the Dawid-Skene algorithm, a machine-learning method used commonly for crowdsourcing to handle label noise. The algorithm takes as input a list of situations, each having a set of noisy “acceptable” and “unacceptable” labels. Then it aggregates all the data and uses some probability calculations to identify patterns in the labels of predicted blind spots and patterns for predicted safe situations. Using that information, it outputs a single aggregated “safe” or “blind spot” label for each situation along with a its confidence level in that label. Notably, the algorithm can learn in a situation where it may have, for instance, performed acceptably 90 percent of the time, the situation is still ambiguous enough to merit a “blind spot.”

In the end, the algorithm produces a type of “heat map,” where each situation from the system’s original training is assigned low-to-high probability of being a blind spot for the system.

“When the system is deployed into the real world, it can use this learned model to act more cautiously and intelligently. If the learned model predicts a state to be a blind spot with high probability, the system can query a human for the acceptable action, allowing for safer execution,” Ramakrishnan says.

By Asit K. Biswas, University of Glasgow and Kris Hartley, The Education University of Hong Kong

In the 21st century, governments cannot ignore how changes in technology will affect employment and political stability.

The automation of work – principally through robotics, artificial intelligence (AI) and the Internet of things (IoT), collectively known as the Fourth Industrial Revolution – will provide an unprecedented boost to productivity and profit. It will also threaten the stability of low- and mid-skilled jobs in many developing and middle-income countries.

From labour to automation

Developing countries must begin seriously considering how technological changes will impact labour trends. Technology now looms just as large a disruptive force, if not larger, than the whims of global capital.

China has for decades increased its global contribution to manufacturing value-added goods, now enjoying a competitive position in Apple products, household appliances, and technology. In the process, the country has made historic progress lifting its citizens out of poverty.

China has accomplished this by raising worker productivity through technology and up-skilling (improving or acquiring new skills), and higher wages have predictably followed.

However, this trend is also compelling manufacturers to relocate some low-skill production to Southeast Asia. US-China trade disputes could exacerbate this trend.

Relocation of manufacturing activity has been an economic boon for workers in countries like Vietnam and Indonesia. However, the race among global manufacturers to procure the cheapest labour brings no assurances of long-term growth and prosperity to any one country.

Governments in developing countries must parlay the proceeds of ephemeral labour cost advantages into infrastructure investment, industrial upgrading and worker upskilling. China has done this to better effect than many.

The growth in sophistication and commercial feasibility of robotics, IoT, and other automation technologies will impact jobs at nearly every skill level. More broadly, the fallout from technological advancement may replicate the disruptive geographic shifts in production once resulting from labour cost arbitrage.

Political blowback

After many decades of globalisation, a borderless economy has emerged in which capital and production move freely to locations with the greatest investment returns and lowest cost structures. This has prompted a pattern of global economic restructuring, generating unprecedented growth opportunities for developing countries.

Workers have been rewarded for their personal efforts in education and skill development, while millions have been lifted from poverty.

Given advancements in technology and the associated impact on livelihoods, it is time to consider how the next chapter of global development will play out politically. Automation will be a highly disruptive force by most economic, social, and political measures. Few countries – developed or otherwise – will escape this challenge.

Some Western countries, including the United States, are already experiencing a populist political wave fuelled in part by the economic grievances of workers displaced from once stable, middle-class manufacturing jobs. Similar push-back may erupt in countries already embroiled in nationalist politics, including India.

Growing populations and the automation of work will soon mix to create unemployment crises, with serious implications for domestic political stability.

As education systems flood the employment market with scores of ambitious graduates, one of the greatest challenges governments face is how to generate well-paying jobs.

Further, vulnerable workers will include not only new entrants but also experienced workers, some of whom are continuously and aggressively up-skilling in anticipation of more lucrative employment.

In India, over 1 million people enter the working-age population every month. More than 8 million new jobs are needed each year to maintain current employment levels.

India’s young population is becoming increasingly pessimistic about their employment prospects. Although official statistics are unreliable, as a large percentage of work occurs in the informal sector in positions such domestic workers, coolies, street vendors, and transient positions lacking contracts, indications are that India may be facing the prospect of jobless growth.

Insufficient skill levels in much of the workforce are impeding India’s effort to accelerate growth in high-productivity jobs. Thus, the country’s large-scale manufacturers, both domestically and internationally owned, are turning to robots to ensure consistent, reliable, and efficient production.

Urbanisation also adds to India’s employment challenge. The promise of higher-paying jobs has lured many rural workers into urban areas, but these workers are often illiterate and lack sufficient skills. This was not always a concern, as these workers could find menial factory jobs. Robots are now doing much of the low-skilled work that migrant workers were once hired to do.

Towards a future of stable livelihoods

The lingering socio-economic imperative for many governments is to replace eliminated jobs. According to The World Economic Forum, “inequality represents the greatest societal concern associated with the Fourth Industrial Revolution.”

However, the WEF and others have given little useful guidance on how to address this challenge. How should the economy absorb multitudes of variously skilled workers displaced by technology?

People aspire to economic and social mobility more than ever before, particularly as they observe wealth rising ostentatiously all around them – on the streets, in the news, and among seemingly lucky friends and acquaintances. Sadly, the aspirations of most will go unfulfilled.

One way forward is said to be through up-skilling by retraining workers to operate and maintain technology systems. However, this seems to be a paradox, as workers would be training robots to eventually take jobs held by humans. If a major driver of automation is reduction or elimination of labour costs, one cannot expect all displaced workers to enjoy stable and continuing employment opportunities.

Despite political promises about employment growth from high-tech industries and the technological transformation of primary sectors, the tension between the drive for technology-based efficiency and the loss of jobs is undeniable and may have no clear resolution.

Societies have reacted to global economic restructuring in discouraging ways, indulging in nationalism, racism, militarism, and arbitrary economic protectionism. Populist opportunists and foul-tempered troglodytes have ridden reactionary rhetoric into positions of political power, raging against what former White House chief strategist Steve Bannon calls the “liberal postwar international order.” At the same time, left-leaning solutions such as universal basic income face significant fiscal and political headwinds.

The 21st century will see increased disruptions to once-stable work life, due to technological progress and the continuing liberalisation of global capital and production. Early indications about how countries will respond – haphazardly and with no clear long-term strategy – are not encouraging.

Asit K. Biswas, Visiting professor, University of Glasgow and Kris Hartley, Assistant professor, The Education University of Hong Kong

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Researchers at IIT-Istituto Italiano di Tecnologia created the first soft robot mimicking plant tendrils. It is able to curl and climb using the same physical principles determining water transport in plants. The research team is led by Barbara Mazzolai, and results have been published in Nature Communications. In the future, this tendril-like soft robot could inspire the development of wearable devices such as soft braces that actively morph their shape.

Semantic segmentation with deep learning

Booth #4220 - We will be showing our HRC-03 grippers for collaborative robotic applications, also our new aluminum guided variant of our very successful 5000 series grippers.

Booth #703 - Cobot Pump, NEW vacuum generator that does not need compressed air, Cobot Gripper, employee and environment safe collaborative gripper and more

Scientists at the Fraunhofer Institute for Machine Tools and Forming Technology IWU have developed an innovative technology enabling people and large industrial robots to work together in an intuitive way that feels a lot like human teamwork. Using the benefit of this technology, robots can recognize gestures, faces and postures to make this collaboration that much safer and more efficient. Fraunhofer IWU is set to present this innovation at the Hannover Messe Preview in hall 19 on January 24, 2019, and at the Hannover Messe in hall 17 at booth C24 from April 1 through 5, 2019.

Amazon on Wednesday began putting cooler-sized robots to work delivering packages to customers in a neighborhood outside Seattle.

In the 21st century, governments cannot ignore how changes in technology will affect employment and political stability.

The profitable companies of the future will be those that can adapt to the change in sourcing, production and distribution that are happening today, and are flexible enough to take advantage of new technologies.

Though often overlooked, grippers are a significant cog in a successful automation system and choosing the proper one can result in optimized performance, uptime and operator safety.

A bipedal robot walks on the surface of a distant planet. Its sensors scan the environment, sending data back to Earth. It comes to a crumbling hillside, flexes its knees and leaps, thrusters whirring, to land safely above the loose rock.

In Europe, people are more reserved regarding robots than they were five years ago. This is shown in a new study published by scientists from Linz and Würzburg.