Headphones have the ability to seal us in our own isolated sound bubbles; putting an invisible wall around wearers, even in public spaces. At least, it can feel that way. In reality, while the world might seem like it disappears when you put on your fancy AirPods Pro, it doesn’t actually. As walking across a busy street without paying attention would quickly remind you.
Could machine intelligence help where human intelligence fails us?
That’s certainly what researchers from Columbia University hope. They have developed a Pedestrian Audio Warning System (PAWS) that seeks to alert headphone wearers of the threat posed by passing vehicles. The smart headphone tech technology works by using machine learning algorithms to interpret vehicle sounds from up to 60 meters away. It can then provide information about the location of those vehicles. The results could be a major boon to pedestrian safety at a time when, tragically, more pedestrians than ever are killed on roads in the United States.
The headphones used for the prototype system feature an array of low-cost microphones, located in different parts of the headset. The relevant sound features of possible cars are extracted by an onboard custom integrated circuit, which then sends them to a paired smartphone app. The smartphone uses machine learning algorithms to determine what is and is not a vehicle sound. The neural network it relies on was trained using audio from a wide range of both vehicles and environmental conditions.
The system is still far from complete. For one thing, it can only identify the approximate position of vehicles; not their trajectory. Being able to determine this would be far more useful than simply assuming a static road state for vehicles that are, in reality, anything but static. Secondly, the researchers are still trying to figure out the best way to signal this information to wearers. One possibility would be to offer warning beeps on different sides of stereo headphones to make clear exactly where a sound is emanating from.
The PAWS project has already received a grant of $1.2 million from the National Science Foundation. According to IEEE Spectrum, the team is hoping to pass a “more refined” version of the tech over to a company that could bring it to market.
San Francisco startup Dimensional advertised for people to walk robot dogs through city streets. The New York Times reported that the job could include carrying a roughly 33-pound robot back to the office if it overheated. A lift home is an awkward requirement for a labor-saving machine.
Keeping the hardware running is only one challenge. Anthropic's robotics tests found that language models mostly failed when controlling joints directly, but did better when directing trained controllers. Giving an AI agent a body doesn't teach it to handle ordinary household objects.
Soon, a mattress protector may be more than a bed cover. Australian medtech company Sleeptite and researchers at RMIT University have received a US patent for FlexiSense, a smart bedding technology that tracks body position and posture to help caregivers identify when older residents may be at risk of falling.
The technology is being developed for Sleeptite’s in-bed monitoring system, REMi, and was recently on display at The MedTech Conference in Boston. A peer-reviewed study in Flexible and Printed Electronics says the sensors can also be made at scale without sacrificing performance or durability.
AI companies spend enormous amounts of time convincing us their technology is safe and secure. But what happens when those same companies become targets of increasingly sophisticated AI-powered attacks? Anthropic, the company behind Claude, is now facing that question. According to a Washington Post report, cybersecurity startup Cathedral discovered vulnerabilities that let its researchers access sensitive information in Anthropic's internal systems.
The researchers used AI agents to identify potential entry points in less than 12 hours. Interestingly, they didn't need to crack Claude itself. Instead, they exploited weaknesses in third-party business applications that Anthropic uses for everyday operations. Anthropic's security testing program disclosed the findings, which allows outside researchers to identify and report vulnerabilities.