Add BrightSurf on Google Email

How mistakes help us recognize things

Researchers at Goethe University Frankfurt found that mistakes in short-term memory, such as misperceiving the motion direction of dots, contribute to our ability to recognize and integrate visual information over time. This 'blurring' of perception helps us perceive a stable environment despite constant changes.

SourceGoethe University Frankfurt·JournalNature Communications·DateApr 28, 2020

Car congestion outweighs scooter scourge on city streets

A recent study by Cornell University researchers found that motor vehicles are the main cause of blocking access to other travelers on city streets, with a higher noncompliance rate (24.7%) compared to scooters and bikes (0.8%). The study suggests that cities should rethink their parking policies in response to the rise of technology-e...

SourceCornell University·JournalTransportation Research Interdisciplinary Perspectives·DateMar 4, 2020

Autonomous vehicles could benefit health if cars are electric and shared

The study found that autonomous vehicles can reduce road accidents and promote physical activity, but the impact depends on factors like technology implementation and fuel use. If implemented correctly, autonomous vehicles could have a significant positive impact on public health, especially if used in electric and shared formats.

SourceBarcelona Institute for Global Health (ISGlobal)·JournalAnnual Review of Public Health·DateJan 30, 2020

Cultural difference play crucial role in when people would sacrifice one to save group

A new study found that people in Eastern countries are less likely to support sacrificing someone to save more lives, highlighting the importance of considering local ethics in AI development. The 'Trolley Dilemma' thought experiment showed a significant difference in willingness to sacrifice one person versus five.

SourceUniversity of Exeter·JournalProceedings of the National Academy of Sciences·DateJan 21, 2020

Deep learning enables real-time imaging around corners

Researchers developed a new laser-based system that can image around corners in real time, enabling applications such as detecting hazards or pedestrians in self-driving cars. The system uses deep learning to reconstruct hidden objects at high resolutions and speeds.

SourceOptica·JournalOptica·DateJan 16, 2020

Researchers bring gaming to autonomous vehicles

A new study created three games for level three and higher semi-autonomous vehicles, which can play with other players nearby. The researchers evaluated the games using a virtual reality driving simulator and participant feedback, finding that participants rated the games highly in immersion and enjoyed playing with strangers.

Teaching cars to drive with foresight

Scientists at the University of Bonn develop an algorithm that completes and interprets LiDAR scans to enable self-driving cars to anticipate potential hazards. The system uses a dataset of superimposed point clouds to improve scene understanding, which can lead to significant improvements in autonomous driving safety.

Driverless cars could lead to more traffic congestion

New research from the University of Adelaide predicts that driverless cars could increase traffic congestion in Australia's City of Adelaide. The study, which surveyed over 500 commuters, found that autonomous vehicles may lead to an adverse impact on public transport, causing peak period vehicle flows to rise.

SourceUniversity of Adelaide·JournalUrban Policy and Research·DateOct 23, 2019

Research reveals new plan to maximize rideshare availability by routing empty cars

Researchers develop a model to control empty car flow in geographic locations to optimize system-wide functionality, directing cars to different locations instead of waiting at the same spot. The algorithm calculates availability of empty cars when requested and finds new passengers quickly at destinations.

Stanford researchers teach robots what humans want

Researchers at Stanford University developed a new system that combines demonstrations and user preference surveys to set goals for autonomous systems, achieving better results in simulations and real-world experiments. The system improved upon previous methods, reducing the time required to generate instructions by 15-50 times.

Spotting objects amid clutter

Researchers at MIT develop a new algorithm that can accurately pick out an object, such as a small animal, in a dense cloud of dots within seconds. The technique prunes away outliers quickly, even for increasingly dense clouds, making it suitable for applications like driverless cars and robotic assistants.