New Year's Robot - TechCrunch
This is always the case weird week - that limited space between the Christmas holidays and the New Year. Romjul - or "Dead Week" - as it is called in Norway (thanks Haje). It's a time for relaxation and reflection on that year for some - and a CES embargo for others. We are currently focusing on the latter, and are trying to be aware of the one that was here at Actuator Headquarters.
We spent much of the past few weeks completing some of the year's major space trends: delivery, warehouse / performance and food preparation. We’ve also spoken to several key figures in the industry, including CMU robotics head Matthew Johnson-Roberson and Boston Dynamics CEO Rob Playter, OpenRobotics CEO Brian Gerkey and iRobot CEO Colin Angle.
This week we have asked MAIL CSAIL Director Daniela Rus to stress the issue. She’s actually surpassed her responses for this last Actuator in 2022, so I’m going to let her put things off.
What was the defining trend of robotics / AI / automation in 2022? In terms of robotics and automation, the pandemic and the subsequent labor shortage made it very clear that robots play a vital role in the workplace. Industry has seen more robots for manufacturing and logistics applications, where independence can deliver value, but independence on the roads in the form of robotaxi is still a long way off. Research has made great strides in making robots safer and more efficient, with advances in the bodies of soft robots and brains of robots with machine learning.
In AI, we have seen a heightened awareness of the challenges with AI solutions today. Industry has adopted many deep cloud network applications that enable devices to enhance work in a number of areas. But it has also become clear that these methods require data accessibility, resulting in very large datasets that need to be manually identified and not readily available in all domains. The quality of that data must be very high, and if the data is biased or bad, the performance of the systems trained on this data will be just as poor. Moreover, these systems are black boxes - there is no way for users of the systems to actually "learn" anything based on internal AI work. There are also strength issues as the trained models are often unstable, and we need to understand that the systems do not do “deep reasoning”, they usually match a shallow pattern. The research community is working to address these challenges.
What will 2022 bring to those regions? As we move into our rapidly changing world, I believe that robots and AI can help us unlock our human potential, as individuals and as a group. While the past 60 years have defined the range of industrial robots and powered hard-to-power machines to perform complex assembly tasks in limited industrial conditions, the next sixty years will be using robots in human environments to help humans with physical activities.
While the industrial robots of the last 60 years have been largely inspired by human form, the next step is soft robots inspired by the animal kingdom: shape and diversity modeled by our own built environment, with the broad potential of being a statement of our natural state. While industrial robots in the last 60 years are made of hard plastic and metal, I believe that the next 60 years will bring machines that are made for us naturally, or through engineering processes such as wood, plastic, paper, ice or even food.
We will also see new ideas for AI, becoming more risky about AI and privacy, and addressing AI sustainability. It is important to remember that the greatest advances today are the result of decades - old observations augmented by a lot of data and calculations. New technical ideas and funding are needed to support them to ensure progress. In addition, we will more clearly understand the carbon footprint of machine learning and become more at risk of Sustainable AI.
AI innovations can help do much of our activity to reduce the impact of warming: optimizing the cost of electricity technology, making transportation more efficient, monitoring and stopping de- afforestation, conserving biodiversity or ensuring there is enough food to go around. But to do that, AI systems need to expend a lot of energy. Current research estimates that deep learning model training produces 626,000 pounds of carbon dioxide, equivalent to five car life emissions. We need to develop simpler models, which can significantly reduce AI 's carbon footprint.
The proliferation of robots, automation and AI technologies has the potential to make people's lives easier - but many of the roles that these technologies can do will detract from the work that humans have done today. We will also focus on efforts to address and respond to potential economic inequalities.
It's impossible to put Professor Russia's words up this week, but I would like to end the year with a few things that clarify what has happened in the industry over the past year. gone and - perhaps - giving us a glimpse of where things are going. I hear what many people have already said in the pages of Actuator: It's been an amazing year for robots. After spending many years covering the space and hearing people finally prognosticate on robots, the pandemic has set in motion much faster than many expected.
Of course, I do not think that anyone was hoping for a pandemic as we got here, but, well, that is the way it will be. And in response to Professor Russia's statement, we hope that we can address concerns about climate change before it is too late - and that we can address the inevitable shift of human jobs. If we are going to use robots to improve the working conditions of some, it is up to us as a society to make sure that we can support those who have jobs that are used as a result.
Robots will tell you about the grim, dirty and dangerous jobs they are planning to replace, but there's a much harder conversation about the effects of moving on people. It's true right now that companies are having a hard time filling vacancies - something that automation can and will face. It is also true that most of this technology is still at a collaborative stage that requires human workers to engage. But as technology advances and becomes more capable, is it unbelievable that we are going to leave the public with great oaths because we believe that the work they do unskilled?
Looking at the trends for the past year, these are the key areas I have been following in the area:
- Product / performance
- Food delivery / prep
- Farming / agriculture
- Medical / surgical
And in some ways, that's just the beginning. For example, a couple of weeks ago we covered Petra, a drilling company that raised $ 30 million to go through hard rock for infrastructure projects. The list of robotics fields is expected to add long and growing. As we look forward to 2022, we will consider how we handle such things, going forward. I am excited to continue discussing these and other issues in Actuator going forward. Next week, though, we get a lot more myopic and focus on CES robots.
Fortunately, the big news about robots has slowed down this Romjul (I can only use my new word once a year, so I make the most of it) . Rita has a good story for you about Meituan, which brings drone - based food delivery to Shenzhen. The Tencent-backed company has delivered 19,000 meals to 8,000 customers in their two-year pilot. As the piece points out, somewhat lax rules in the manufacturing hub are a big part of what brought it to Shenzhen.
Speaking of China, the country’s Ministry of Industry and Information Technology has helped map out an ambitious five-year plan aimed at increasing robot adoption. That includes a planned 20% increase in revenue per year, with a focus on manufacturing.
Happy New Year, and congratulations on getting through this last one. Not everyone was so lucky. Here's to share more robot news in 2022.
Need a head start on your New Year’s resolution by subscribing to more free robotics newsletters? Have good news for you.