If a robot traveling to a destination has just two possible paths, it needs only to compare the routes’ travel time and probability of success. But if the robot is traversing a complex environment with many possible paths, choosing the best route amid so much uncertainty can quickly become an intractable problem.

MIT researchers developed a method that could help this robot efficiently reason about the best routes to its destination. They created an algorithm for constructing roadmaps of an uncertain environment that balances the tradeoff between roadmap quality and computational efficiency, enabling the robot to quickly find a traversable route that minimizes travel time.

The algorithm starts with paths that are certain to be safe and automatically finds shortcuts the robot could take to reduce the overall travel time. In simulated experiments, the researchers found that their algorithm can achieve a better balance between planning performance and efficiency in comparison to other baselines, which prioritize one or the other.

This algorithm could have applications in areas like exploration, perhaps by helping a robot plan the best way to travel to the edge of a distant crater across the uneven surface of Mars. It could also aid a search-and-rescue drone in finding the quickest route to someone stranded on a remote mountainside.

“It is unrealistic, especially in very large outdoor environments, that you would know exactly where you can and can’t traverse. But if we have just a little bit of information about our environment, we can use that to build a high-quality roadmap,” says Yasmin Veys, an electrical engineering and computer science (EECS) graduate student and lead author of a paper on this technique.

Veys wrote the paper with Martina Stadler Kurtz, a graduate student in the MIT Department of Aeronautics and Astronautics, and senior author Nicholas Roy, an MIT professor of aeronautics and astronautics and a member of the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). The research will be presented at the International Conference on Robotics and Automation.

Generating graphs

To study motion planning, researchers often think about a robot’s environment like a graph, where a series of “edges,” or line segments, represent possible paths between a starting point and a goal.

Veys and her collaborators used a graph representation called the Canadian Traveler’s Problem (CTP), which draws its name from frustrated Canadian motorists who must turn back and find a new route when the road ahead is blocked by snow.

In a CTP, each edge of the graph has a weight associated with it, which represents how long that path will take to traverse, and a probability of how likely it is to be traversable. The goal in a CTP is to minimize travel time to the destination.

The researchers focused on how to automatically generate a CTP graph that effectively represents an uncertain environment.

“If we are navigating in an environment, it is possible that we have some information, so we are not just going in blind. While it isn’t a detailed navigation plan, it gives us a sense of what we are working with. The crux of this work is trying to capture that within the CTP graph,” adds Kurtz.

Their algorithm assumes this partial information — perhaps a satellite image — can be divided into specific areas (a lake might be one area, an open field another, etc.)

Each area has a probability that the robot can travel across it. For instance, it is more likely a nonaquatic robot can drive across a field than through a lake, so the probability for a field would be higher.

The algorithm uses this information to build an initial graph through open space, mapping out a conservative path that is slow but definitely traversable. Then it uses a metric the team developed to determine which edges, or shortcut paths through uncertain regions, should be added to the graph to cut down on the overall travel time.

Selecting shortcuts

By only selecting shortcuts that are likely to be traversable, the algorithm keeps the planning process from becoming needlessly complicated.

“The quality of the motion plan is dependent on the quality of graph. If that graph doesn’t have good paths in it, then the algorithm can’t give you a good plan,” Veys explains.

After testing the algorithm in more than 100 simulated experiments with increasingly complex environments, the researchers found that it could consistently outperform baseline methods that don’t consider probabilities. They also tested it using an aerial campus map of MIT to show that it could be effective in real-world, urban environments.

In the future, they want to enhance the algorithm so it can work in more than two dimensions, which could enable its use for complicated robotic manipulation problems. They are also interested in studying the mismatch between CTP graphs and the real-world environments those graphs represent.

“Robots that operate in the real world are plagued by uncertainty, whether in the available sensor data, prior knowledge about the environment, or about how other agents will behave. Unfortunately, dealing with these uncertainties incurs a high computational cost,” says Seth Hutchinson, professor and KUKA Chair for Robotics in the School of Interactive Computing at Georgia Tech, who was not involved with this research. “This work addresses these issues by proposing a clever approximation scheme that can be used to efficiently compute uncertainty-tolerant plans.”

This research was funded, in part, by the U.S. Army Research Labs under the Distributed Collaborative Intelligent Systems and Technologies Collaborative Research Alliance and by the Joseph T. Corso and Lily Corso Graduate Fellowship.

Ubiquity is Everything in the AI Chatbot World

Companies have raced to build desktop and mobile apps for their bots, in order to both give them new capabilities but also to make sure they’re right in front of your face as often as possible.

Google’s New Gemini App

The free app is simple and straightforward: it’s just a chat window and a list of your previous chats. You can query the bot with text, voice, or your camera, and it’ll give you answers. It’s effectively identical to the Gemini section of the Google app, or what you’d get by opening a browser and going to the Gemini website.

New Features and Access

The Gemini app does have one newish feature: access to Gemini Live, the bot’s more interactive and conversational chat mode that is similar to ChatGPT’s voice mode. Gemini Live has been available on Android for a few weeks, but this is the first place it has been usable for iPhone owners.

Why Ubiquity is Key

But Live will eventually be everywhere. Whenever the next version of Gemini comes out, that will be too. The whole point of the Gemini app is to put the icon on your homescreen, and give you something to assign to the Action Button or one of the other quick-access spots on your phone.

Conclusion

The Gemini app is just another example of how companies are racing to put their bots in front of our faces as often as possible. With the ability to query the bot with text, voice, or camera, and access to other Google apps, the Gemini app is a step in the right direction. But as the race for our homescreens continues, it remains to be seen whether users will make the bot a habit.

FAQs

Q: What is the Gemini app?
A: The Gemini app is a free chatbot app that allows you to query the bot with text, voice, or camera, and access other Google apps.

Q: What is Gemini Live?
A: Gemini Live is the bot’s more interactive and conversational chat mode that is similar to ChatGPT’s voice mode.

Q: Why is ubiquity important for AI chatbots?
A: Ubiquity is important for AI chatbots because it allows users to interact with the bot frequently, building muscle memory and increasing the chances of making it a habit.

MIT class 2.679 (Electronics for Mechanical Systems II) offers a sort of alchemy that transforms students from consumers of knowledge to explorers and innovators, and equips them with a range of important new tools at their disposal, students say.

“Topics which could otherwise feel intimidating are well-scoped each week so that students come out knowing not only what a concept is, but why it’s useful and how to actually implement it,” says graduating senior Audrey Chen. “I could consistently come in with no background and come out with practical experience I could use in future projects. I’d describe the class as a series of small crash courses [each of which] answers, simply, ‘what do I need to know to do or use this thing?’”

The course takes students through the process of design, fabrication, and assembly of a printed circuit board (PCB). Ultimately, that process, which has twists and turns depending on each student’s project idea, culminates in incorporating the PCB into a device — in a sense animating that device to perform a certain function.

“The design intent of 2.679 is to empower students to ‘imagine it, build it,’” says Tonio Buonassisi, professor of mechanical engineering. “Between those two is a universe, and the purpose of this class is to aid aspiring engineers to bridge that gap.”

Senior Jessica Lam marvels at how much she learned in the course over its one short semester, attributing that flood of education to the class labs being “incredibly well-structured.”

“I’ve found that in a lot of other labs and project-based classes, they throw a lot of information at you at once with the expectation that you already have some experience with certain software or hardware, and most of it is scaffolded and feels like a black box,” without much understanding of what is actually happening, Lam says. “In 2.679, Steve Banzaert has a better understanding of what we already know and how to build on that.”

After taking 2.679, she says she feels “a lot more confident in designing electrical systems, and I have a more comprehensive understanding of how to integrate mechanical systems and electronics.”

Banzaert, technical instructor for the course, says the class is designed to guide students along their own chosen paths of discovery, showing them that they are able to address the challenges they encounter along the way.

“Every semester we get to see really lovely examples of growth, not just in the course material but, in the best cases, in students’ understanding of what they’re really capable of,” he says.

Chen, a mechanical engineering major who is graduating early to start a position as a hardware project manager at Formlabs, agrees that the class did just that.

“Students are given tremendous freedom to pick their own final projects, allowing them to explore topics which are of special interest to them. And because each project is unique, there is less pressure to ‘perform’ in a traditional sense,” she says. “Rather, each student is learning different skills and is encouraged to get as far along with the project they choose as possible. Steve emphasized that the scope of our projects would inevitably change, because at the start you simply don’t yet know what you don’t know, and that’s totally okay!”

Banzaert says, “We try to make it very clear that, yes, we are talking about important general concepts in theory and analysis, but that’s because they are tools that engineers use to solve problems. I think maybe this focus helps remind the students of what got them here in the first place — that the reason you’re an engineer is because there’s something about the world you wish was better, that you’re the person to do it (or at least help), and, if you want to do it well, you’re going to have to learn a bunch of things so you have more tools in your toolbox.”

Senior Yasin Hamed designed a car in the class that uses computer vision to follow along a black line. The car has an attached camera that captures images and relays them to a Raspberry Pi computer that is also attached to the car. Processing the images in real time allows the car to locate the black line and turn or go straight while controlling the car’s speed.

Although Hamed, who is majoring in mechanical engineering with a minor in computer science, had built another similar system in a previous class, he says the focus in the prior class was on the software. With his 2.679 car project, he learned about “the underlying foundation,” meaning “the design of the power electronics and control circuitry which is necessary for everything else to work.”

“I derived much of the ‘enlightenment’ from this class from the little electronic bits and pieces of information I picked up along the course of the class, like learning/practicing soldering, understand how to use integrated circuits, learning how to design a PCB, etc.,” he says. “It was the collection of all of these things that benefited me the most.”

Jordan Parker-Ashe, also a senior, appreciated how 2.679 combined lessons about electronics with research and presentations from Buonassisi’s lab. “It’s great seeing engineering applied in research,” she says.

Although many of the skills she learned in the course were new to her, one was “an old foe,” she says, that 2.679 allowed her to befriend. Parker-Ashe, who is majoring in nuclear engineering, had used a computer vision program called OpenCV in her first Undergraduate Research Opportunities Program project as a first-year undergraduate.

“It was the hardest thing ever, and it really felt like an insurmountable obstacle then,” she says. “Now, to be using OpenCV in labs and homework effortlessly — It was a very full-circle moment.”

She says the class has opened up a whole new field to her, with Banzaert having “directly inspired” her to also take class 6.131 (Power Electronics), “which has been life-changing,” she says.

“2.679 helped me believe in myself, which inspired me to take 6.131, a notorious electrical engineering capstone, which has made me realize that my future lies as a nuclear-electrical engineering engineer, not just a nuclear engineer,” Parker-Ashe says. “I want to pursue electrical engineering in my future, and that just wasn’t on the table beforehand.

“Not to mention that it’s opened the doors to very rich landscapes for project ideas, creating explorations, art, stepping into new roles in group projects, etc,” she says. “I’m so glad that I’ve been able to find opportunities in Course 2  that helped give me hands-on, applied engineering experience.”

OpenAI Expands into Robotics

OpenAI is evidentally ramping up its own robotics efforts, too. Last week, Caitlin Kalinowski, who previously led the development of virtual and augmented reality headsets at Meta, announced on LinkedIn that she was joining OpenAI to work on hardware, including robotics.

Partnership with Physical Intelligence

Lachy Groom, a friend of OpenAI CEO Sam Altman and an investor and cofounder of Physical Intelligence, joins the team at the conference room to discuss the business side of the plan. Groom wears an expensive-looking hoodie and seems remarkably young. He stresses that Physical Intelligence has plenty of runway to pursue a breakthrough in robot learning. “I just had a call with Kushner,” he says in reference to Joshua Kushner, founder and managing partner of Thrive Capital, which led the startup’s seed investment round. He’s also, of course, the brother of Donald Trump’s son-in-law Jared Kushner.

Other Companies Chasing the Same Breakthrough

A few other companies are now chasing the same kind of breakthrough. One called Skild, founded by roboticists from Carnegie Mellon University, raised $300 million in July. “Just as OpenAI built ChatGPT for language, we are building a general purpose brain for robots,” says Deepak Pathak, Skild’s CEO and an assistant professor at CMU.

Challenges in Achieving a Breakthrough

Not everyone is sure that this can be achieved in the same way that OpenAI cracked AI’s language code.

There is simply no internet-scale repository of robot actions similar to the text and image data available for training LLMs. Achieving a breakthrough in physical intelligence might require exponentially more data anyway.

“Words in sequence are, dimensionally speaking, a tiny little toy compared to all the motion and activity of objects in the physical world,” says Illah Nourbakhsh, a roboticist at CMU who is not involved with Skild. “The degrees of freedom we have in the physical world are so much more than just the letters in the alphabet.”

Cautions from Experts

Ken Goldberg, an academic at UC Berkeley who works on applying AI to robots, cautions that the excitement building around the idea of a data-powered robot revolution as well as humanoids is reaching hype-like proportions. “To reach expected performance levels, we’ll need ‘good old-fashioned engineering,’ modularity, algorithms, and metrics,” he says.

Russ Tedrake, a computer scientist at the Massachusetts Institute of Technology and vice president of robotics research at Toyota Research Institute says the success of LLMs has caused many roboticists, himself included, to rethink his research priorities and focus on finding ways to pursue robotic learning on a more ambitious scale. But he admits that formidable challenges remain.

Conclusion

While OpenAI is making significant strides in the field of robotics, the path to achieving a breakthrough in physical intelligence is likely to be long and challenging. The lack of internet-scale data and the complexity of the physical world pose significant hurdles. However, the potential rewards of successfully developing robots that can learn and adapt in the physical world are substantial.

FAQs

Q: What is OpenAI’s plan for robotics?

A: OpenAI is ramping up its robotics efforts, with Caitlin Kalinowski joining the team to work on hardware, including robotics.

Q: Who is Lachy Groom?

A: Lachy Groom is a friend of OpenAI CEO Sam Altman and an investor and cofounder of Physical Intelligence.

Q: What is Physical Intelligence?

A: Physical Intelligence is a startup that is working on developing robots that can learn and adapt in the physical world.

Q: What is Skild?

A: Skild is a startup founded by roboticists from Carnegie Mellon University that is working on developing a general-purpose brain for robots.

Q: What are the challenges in achieving a breakthrough in physical intelligence?

A: The lack of internet-scale data and the complexity of the physical world pose significant hurdles in achieving a breakthrough in physical intelligence.

Robotics and automation are everywhere these days and they are helping all kinds of companies, no matter their size. Whether they be big or small, businesses are improving and reaching new heights because of the helping hands offered by this cutting-edge, top-of-the-line technology.

If you run a small business, you are missing out on a lot by not investing in and using robotics and automation. It can make a huge difference and take your profits to the next level.

Fast Invoicing

Invoicing is a vital part of most businesses but, honestly, it can be a bit of a bore too. Sadly, it takes a lot of time, which means an employee has to devote many hours to making sure that customers, businesses, and other parties are invoiced correctly.

But automation has led to the creation of simple invoicing, which can all be handled by modern apps that get the job done in seconds, instead of hours. Obviously, the business owner has to load the proper data, such as prices, names, dates, and more.

But once it is all put in the system, a simple invoicing program will handle the rest and ensure that everyone is delivered the invoices when they need them, perfectly on time.

By saving time, businesses can ensure their employees are tackling another task that simply must be done by a human.

Improved Workflow

Yet another way that small businesses improve because of automation is by enhancing their everyday workflows, boosting productivity and enhancing morale too. Automation tools can handle repetitive, time-consuming tasks that previously needed manual involvement.

Think about administrative duties – things like scheduling, invoicing, entering data, and managing emails. When they automate these tasks, small business owners can once again claim their time, which enables them to instead focus on more important and strategic endeavors.

For example, inventory control is a subject in which automation can have an important and positive effect. Small businesses often run into difficulties in manually tracking inventory levels, reordering materials, or overseeing turnover.

Automated inventory management systems allow businesses to effortlessly monitor stock levels in real-time and even automatically submit orders when inventory is low.

This can eliminate the likelihood of overstocking or depleting inventory, preventing expensive mistakes and enhancing cash flow.

Warehouse

For small companies that handle certain goods, logistics and fulfillment can become a total mess and a headache. But this is another area where automation and robotics can play a helpful role.

Automated systems and robots in warehouses can perform important and specific tasks such as picking out items, sorting them, packing them up, and then shipping them as well. Now, these are jobs that would usually require a lot of physical work from human employees.

Now think of how robots and automation can help with this situation. Small companies that run e-commerce sites can rely on robotics to streamline their order fulfillment process.

Meanwhile, robots are capable of choosing products from shelves, packaging them, and getting them ready for shipping.

Not only does this save physical labor, but it is also done much faster than it would be by humans, while also reducing big errors such as missing or wrong items.

On top of all that, using robotic systems in warehouses can help companies by optimizing their space utilization.

Robots can move through tight and tiny aisles as well as manage inventory effectively, which obviously allows the optimization of precious space that can be used for more stock or customer-centered activities.

Reducing Labor

All business owners know that labor represents a huge cost, especially for tasks that involve a large amount of manual labor.

It doesn’t matter if it’s packaging products, organizing them, or keeping the workplace clean, these kinds of activities take a lot of time as well as a lot of money when they are carried out by humans.

As you can imagine, robotics helps with all of that. Putting robots into these activities lets small businesses lessen their need for manual labor, lower their overtime expenses, and decrease the need for temp staff during busy seasons.

Robots do not ask for breaks, they don’t require sick leave and are able to work for long periods of time without any sort of decrease in productivity. These are all benefits that turn them into a great choice for human employees.

On top of all that, robots also offer precision and consistency, which lowers the chance of costly mistakes.

When it comes to areas such as manufacturing, product quality is crucial and the use of robotics and automation will result in fewer mistakes, which will definitely save a lot of time, energy, and money.

Conclusion

No matter what you think, robotics and automation are here to stay. And they can help small businesses in so many ways. In fact, it’s impossible to find a single part of a business that cannot be at least somewhat improved by these two parts of modern technology.