The popular Japanese automaker, Toyota is currently working with MIT (Massachusetts Institute of Technology) to create improved systems that can make self-driving cars perceive and respond to any driving environment just like humans do. This type of research is geared towards giving autonomous vehicles the ability to recognize patterns and also learn from previous data in order to find its way around an unpredictable or new environment.
Toyota is currently working with MIT to come up with improved systems for self-driving vehicles of the future
According to the latest reports, Toyota and MIT recently released “DriveSeg” – a new, improved and open dataset based on their research. This DriveSeg as seen in the picture above is created to visibly demonstrate how future self-driving systems could actually perceive visual information from its immediate environment in a non-ending flow.
Bryan Reimer, the principal researcher at MIT said in a recent public statement that DriveSeg was made available to the public for free because the team is hoping to see contributions from other researchers. Reports have it that the dataset in DriveSeg shares close resemblance to real-world dynamic driving situations and this will allow researchers to easily explore how different patterns can play out over time.
Below is a high-resolution video demonstrating the manual version of DriveSeg;
MIT DriveSeg Dataset for Dynamic Driving Scene Segmentation Demonstration Video
This DriveSeg dataset is currently made up of 2 parts: the first being a 3-minute high-resolution video which is featured above, while the second part is a Semi-Auto version that basically represents a total of 20,100 video frames that is taken from MIT’s (AVT) Consortium data.
While experts are still asking if Nigeria is ready for Self-driving cars or not because of so many unresolved challenges that could come up, this new research by Toyota and MIT could shine a new light of hope.