At the top of many automation wish lists is a particularly time-consuming task: chores.

The moonshot of many roboticists is cooking up the proper hardware and software combination so that a machine can learn “generalist” policies (the rules and strategies that guide robot behavior) that work everywhere, under all conditions.

Realistically, though, if you have a home robot, you probably don’t care much about it working for your neighbors. MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers decided, with that in mind, to attempt to find a solution to easily train robust robot policies for very specific environments.

“We aim for robots to perform exceptionally well under disturbances, distractions, varying lighting conditions, and changes in object poses, all within a single environment,” says Marcel Torne Villasevil, MIT CSAIL research assistant in the Improbable AI lab and lead author on a paper about the work, which appears on the preprint server arXiv.

“We propose a method to create digital twins on the fly using the latest advances in computer vision. With just their phones, anyone can capture a digital replica of the real world, and the robots can train in a simulated environment much faster than the real world, thanks to GPU parallelization. Our approach eliminates the need for extensive reward engineering by leveraging a few real-world demonstrations to jump-start the training process.”

Taking your robot home

RialTo, of course, is a little more complicated than just a simple wave of a phone and (boom!) home bot at your service. It begins by using your device to scan the target environment using tools like NeRFStudio, ARCode, or Polycam. Once the scene is reconstructed, users can upload it to RialTo’s interface to make detailed adjustments, add necessary joints to the robots, and more.

The refined scene is exported and brought into the simulator. Here, the aim is to develop a policy based on real-world actions and observations, such as one for grabbing a cup on a counter. These real-world demonstrations are replicated in the simulation, providing some valuable data for reinforcement learning.

“This helps in creating a strong policy that works well in both the simulation and the real world. An enhanced algorithm using reinforcement learning helps guide this process, to ensure the policy is effective when applied outside of the simulator,” says Torne.

Testing showed that RialTo created strong policies for a variety of tasks, whether in controlled lab settings or more unpredictable real-world environments, improving 67% over imitation learning with the same number of demonstrations. The tasks involved opening a toaster, placing a book on a shelf, putting a plate on a rack, placing a mug on a shelf, opening a drawer, and opening a cabinet.

For each task, the researchers tested the system’s performance under three increasing levels of difficulty: randomizing object poses, adding visual distractors, and applying physical disturbances during task executions. When paired with real-world data, the system outperformed traditional imitation-learning methods, especially in situations with lots of visual distractions or physical disruptions.

“These experiments show that if we care about being very robust to one particular environment, the best idea is to leverage digital twins instead of trying to obtain robustness with large-scale data collection in diverse environments,” says Pulkit Agrawal, director of Improbable AI Lab, MIT electrical engineering and computer science (EECS) associate professor, MIT CSAIL principal investigator, and senior author on the work.

As far as limitations, RialTo currently takes three days to be fully trained. To speed this up, the team mentions improving the underlying algorithms and using foundation models. Training in simulation also has its limitations, and currently it’s difficult to do effortless sim-to-real transfer and simulate deformable objects or liquids.

The next level

So what’s next for RialTo’s journey? Building on previous efforts, the scientists are working on preserving robustness against various disturbances while improving the model’s adaptability to new environments.

“Our next endeavor is this approach to using pre-trained models, accelerating the learning process, minimizing human input, and achieving broader generalization capabilities,” says Torne.

“We’re incredibly enthusiastic about our ‘on-the-fly’ robot programming concept, where robots can autonomously scan their environment and learn how to solve specific tasks in simulation. While our current method has limitations—such as requiring a few initial demonstrations by a human and significant compute time for training these policies (up to three days)—we see it as a significant step towards achieving ‘on-the-fly’ robot learning and deployment,” says Torne.

“This approach moves us closer to a future where robots won’t need a preexisting policy that covers every scenario. Instead, they can rapidly learn new tasks without extensive real-world interaction. In my view, this advancement could expedite the practical application of robotics far sooner than relying solely on a universal, all-encompassing policy.”

“To deploy robots in the real world, researchers have traditionally relied on methods such as imitation learning from expert data, which can be expensive, or reinforcement learning, which can be unsafe,” says Zoey Chen, a computer science Ph.D. student at the University of Washington who wasn’t involved in the paper.

“RialTo directly addresses both the safety constraints of real-world RL [robot learning], and efficient data constraints for data-driven learning methods, with its novel real-to-sim-to-real pipeline.

“This novel pipeline not only ensures safe and robust training in simulation before real-world deployment, but also significantly improves the efficiency of data collection. RialTo has the potential to significantly scale up robot learning and allows robots to adapt to complex real-world scenarios much more effectively.”

“Simulation has shown impressive capabilities on real robots by providing inexpensive, possibly infinite data for policy learning,” adds Marius Memmel, a computer science Ph.D. student at the University of Washington who wasn’t involved in the work.

“However, these methods are limited to a few specific scenarios, and constructing the corresponding simulations is expensive and laborious. RialTo provides an easy-to-use tool to reconstruct real-world environments in minutes instead of hours.

“Furthermore, it makes extensive use of collected demonstrations during policy learning, minimizing the burden on the operator and reducing the sim2real gap. RialTo demonstrates robustness to object poses and disturbances, showing incredible real-world performance without requiring extensive simulator construction and data collection.”

This story is republished courtesy of MIT News (web.mit.edu/newsoffice/), a popular site that covers news about MIT research, innovation and teaching.

Research, conducted by The University of Warwick and the University of New South Wales in Australia, analyzes animal sounds from endangered species including types of elephants, whales and birds.

It uses a new method adapted from tech used to analyze brain waves in neuroscience. The study is published in the journal Ecology and Evolution.

Analysis of animal sounds can be used to estimate their population size, to identify what animals live in a particular area, to understand their migration patterns, and to understand any negative impacts they may experience due to the increasing levels of noise created by human activity that are occurring in most of their habitats. Such insights are vital in developing environmental management and conservation strategies.

The new method was shown to be more accurate than the conventional methods for analysis of animal sounds. While testing this new method, called the Superlet transform, the study also revealed some previously unreported or disputed details in animal sounds:

• The Asian elephant call isn’t just made up of continuous tones, but also contains sounds that are “pulsed,” or comprised of regularly timed bursts of sound energy.
• Pulsing was also shown in the southern cassowary (a large bird similar to an emu) and American crocodile calls.
• New evidence was uncovered that helps to solve a debate around the characteristics of the Chagos pygmy blue whale’s song.

These are not conclusive findings, as each one is based on just a single recording. To confirm them, more sounds will need to be analyzed. They illustrate, however, the power of this new method to clarify details that previously might have been ambiguous.

Lead Researcher Ben Jancovich, a Ph.D. candidate from The University of New South Wales, and visiting Ph.D. student at The University of Warwick’s Mathematics Institute, said, “Our new study highlights that sometimes, the accepted tools that we’ve become comfortable with, may not actually be the best tools for the job.”

“This is especially true in cross-disciplinary fields like bioacoustics, where the methods are highly technical, and require expertise in multiple fields.”

“The new method we demonstrated offers increased accuracy and requires less expertise to use, so it should prove to be a hugely valuable tool for animal sound researchers that don’t have an engineering background.”

Current methods (including the “Short-Time Fourier Transform,” STFT) have difficulties in accurately revealing both the rhythms and pitch of sounds at the same time.

These limitations are more pronounced at lower frequencies, affecting the analysis of sounds like those made by blue whales—gentle giants of the sea, sadly listed as an endangered species.

The new technology will be available for people to use for free, via a simple-to-use app, making it easy for researchers from different fields to use, without needing extensive knowledge of audio signal analysis.

An analysis of how rhinoceros beetles deploy and retract their hindwings shows that the process is passive, requiring no muscular activity. The findings, reported in Nature, could help improve the design of flying micromachines.

Among all flying insects, beetles demonstrate the most complex wing mechanisms, involving two sets of wings: a pair of hardened forewings called elytra and a set of delicate membranous hindwings. Although extensive research exists on the origami-like folds of their wings, little is known about how they deploy and retract their hindwings.

Previous research theorizes that thoracic muscles drive a beetle’s hindwing base movement, but experimental evidence to support this theory is lacking.

Hoang-Vu Phan and colleagues combine the use of high-speed cameras and a dynamically similar flying robot to address this research gap. The authors observe that rhinoceros beetles use passive mechanisms, including their elytra, when deploying and retracting their wings.

Deployment is a two-phase process in which elevation of the elytra partially releases their hindwings in a spring-like fashion, then a flapping motion brings the hindwings into a raised flight position. They also find the beetles use their elytra to lower their hindwings into a resting position passively.

Inspired by their observations, the authors make microrobots that mimic beetle wings’ passive deployment and retraction. They find the bots successfully take off and maintain flight.

Their findings suggest that transferring the beetle’s passive hindwing processes to a flapping robot design could help to improve the capabilities of small robots that need to operate in limited or cluttered spaces.