Population growth and urbanization are increasing traffic congestion and pollution. Cities around the world are implementing high occupancy lanes and congestion charging to promote ride-sharing and public transportation. Working with the Ministry of Transportation in Ontario and private operators, we developed a roadside unit that counts occupants in fast moving vehicles in all weather conditions. The system uses a combination of lidar and near infrared sensors to accurately count occupants in all conditions.
Working with a leading manufacturer of office furnishings we have developed a system to gather information about how office space is used throughout the day. This enables their designers to identify congested and underutilized areas, achieve the right balance of quiet and collaborative space, and create pleasing, healthy, productive work environments. Our software runs directly on inexpensive top view cameras, making installation and management straightforward. Maintaining personal privacy is an essential requirement: no video or still images of any environment are stored, just metadata about what is happening in each area.
We are building an advanced safety system for autonomous rail vehicles working with a world leading provider of technology and control systems for rail networks. The system employs high resolution video and radar sensors together with our software to enhance frontal situational awareness. We have developed a new (patent pending) approach to fusing radar and video data that delivers higher accuracy and performance than existing methods. The system will identify pedestrians, road and rail vehicles, and other obstacles in all weather and lighting conditions and outside a driver’s field of view. The system will be tested by operators in North America and Europe.
Safety & Security
There are millions of security camera installed in public and commercial spaces to enhance security and safety. We are working with a leading global electronics company to replace motion alerts and simple video analytics with advanced AI-based analytics. We are running multi-class object detection, classification and tracking on camera. Networks of thousands of cameras are securely managed by an onsite fog layer, which in turn is securely connected to a private cloud. This architecture minimizes network traffic and ensures that the system at each site remains fully functional if Internet connectivity is lost. Future versions of the system will provide for object tracking across multiple sensors.
As of 2018 rear view cameras are now fitted by law to every new car and light truck sold in North America. Working with a major automotive supplier we have taken these passive video systems to the next level by adding real time collision warning. Our software runs on camera, which was a challenge given the extremely limited memory and processing resources. We detect people and other obstacles, clearly identify them on screen, and alert the drive to potential collisions. Future version of the system can be integrated with parking sensors and braking systems.
We are working with the Canadian government to develop and test systems to help keep our soldiers safe. This includes a combination of fixed, ground vehicle and aerial vehicle mounted sensors. The sensors include high resolution video, thermal, radar and lidar, both standalone and fused with complementary sensors to enhance performance in challenging environmental conditions. The systems enhance situational awareness by identifying threats and anomalies in real time. Many of the advances being made on these military applications are directly applicable to our civilian security and safety applications.