Absent-minded Robots Remember What Matters
We are constantly inundated with new information, and to manage it effectively it’s sometimes necessary to forget old, irrelevant memories.
Researchers at Vanderbilt University have now developed an algorithm that mimics this kind of forgetfulness in robots, as a way to filter out less useful information.
“Forgetting is a critical capability when operating in dynamic environments,” says PhD student Sanford Freedman, who presented the group’s data filtering-software, called ActSimple, in a paper published at the IASTED Robotics and Applications conference held this week in Cambridge, MA.
ActSimple draws on two facets of human memory: time-based decay, or the way that memories disappear over time, and interference, which is the failure to recall information due to other memories competing for attention. ActSimple assigns different pieces of data values depending on how often they are used, and how similar it is to other pieces of information.
To test the software, the researchers used it to control a simulated robot that measured the strength of WiFi signals in a virtual environment. The robot recorded WiFi readings on a scale of 1-100, as it moved through the virtual setting and these WiFi readings also had different levels of noise (errors) associated with them. At intervals, the robot relied on its memory to create an estimated WiFi signal map by recalling signal strength information it had gathered and stored. The researchers tested ActSimple against four other algorithms, including one that strictly disregarded the oldest information, and another that out filtered random information.
The Team found that on average, ActSimple created the most reliable estimated WiFi map. Interestingly, when the robot “remembered” everything–that is, used all of its gathered information (errors and all)–it generated the least accurate map overall.
Keep Reading
Most Popular
Large language models can do jaw-dropping things. But nobody knows exactly why.
And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.
How scientists traced a mysterious covid case back to six toilets
When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.
The problem with plug-in hybrids? Their drivers.
Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.
Google DeepMind’s new generative model makes Super Mario–like games from scratch
Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.
Stay connected
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.