Looking Forward, Seeing Ahead: AI Technology Gives a Clearer View of the Future

Click here if the video doesn't play.

21st May, 2019

Training cameras to detect and recognize objects at great distances has proven a formidable technological challenge. But no obstacle is unconquerable; Mitsubishi Electric has actually succeeded in extending the range of detection to an industry-leading distance standard. We couple our Maisart® AI technology and proprietary algorithm with an electronic mirror and have made it possible to detect objects at distances of up to 100 meters — the length of a soccer pitch. This technology has given us new eyes – and the prospects on the horizon are limitless.

Knowledge Is Safety

Humankind has long applied its critical thinking ability to the development of technologies that improve our vision in various situations, both to ensure our safety and to expand our base of knowledge about the world around us. We at Mitsubishi Electric have taken up the mantle, establishing new industry benchmarks in the process.

One aspect of our lives in which demand for enhanced vision is particularly strong is driving. A significant proportion of automobile accidents are the result of lane changes, caused in many cases by the driver’s inability to sufficiently confirm the presence or absence of vehicles in the so-called "blind spot." Instead of relying on our ability to react instantaneously, we need to be able to anticipate potentially hazardous situations to truly achieve a smarter and safer society. The electronic mirror, a device designed to provide visibility superior to that of rear-view and side-view mirrors, is part of the move to address this issue. Limitations remain, however, as the conventional electronic mirror only offers 30 meters of detection range at a detection accuracy rate of 14 percent. In response, we developed Object Recognition Technology, and the results are astounding.

Eye-opening Innovation

The component of Object Recognition Technology that is vital to its operation is the Visual-Cognition Model. An algorithm that imitates the human tendency to give preferential attention to areas in our field of vision that stand out, the Visual-Cognition Model essentially eliminates all other elements in the field, thereby reducing clutter and greatly accelerating decision making. And by incorporating our proprietary Maisart® AI technology, we have further achieved not only the real-time detection of objects, but also the identification of those objects.

Through the use of Object Recognition Technology with an electronic mirror in an automotive application, we have succeeded in extending the maximum detection range to an amazing 100 meters, and in elevating the detection accuracy rate to 81 percent, both tops in the industry*. And the improvement of convenience and safety on the roads of the world is just the tip of the iceberg for our extraordinary technology.

Please refer to the "News release" link below for more details.

Sights on the Future

We are at work creating value-added functionality by augmenting Object Recognition Technology with advanced sensing capabilities. This will make it possible to more rapidly assess dangerous circumstances and provide the information to the driver. And its value extends well beyond ground transportation. With the prevalence of cameras in today’s society, we see tremendous potential for this technology, extending to capabilities such as identifying objects in poorly lit environments, and ascertaining the velocity of approaching objects. Object Recognition Technology could someday potentially even be applied to other modes of transportation, such as air travel. The possibilities are virtually limitless.

We are capitalizing on our technological mastery to see beyond what our eyes are able to show us, with the objective of spearheading the evolution to a safer, more secure society.

The content is true and accurate as of the time of publication.Information related to products and services included in this article may differ by country or region.

Related Content