Study Compares Human Drivers and Autonomous Vehicle Technologies

 

How do human driving skills — based on flesh-and-blood sensory perception, cognition and eye-hand coordination — measure up against autonomous vehicle technologies?

According to a new study, human drivers — when attentive, sober and well rested — still generally maintain an advantage in terms of reasoning, perception and sensing. But machines and computers are quite capable of performing tasks such as driving and have an edge in the areas of reaction time, power output and control, consistency and multichannel information processing.

The study, released by Sustainable Worldwide Transportation at the University of Michigan, offers a scientific take on the man-versus-machine driving debate. Human eyes, for example, are compared to autonomous vehicle radar, LIDAR and camera systems.

One advantage that people have is that their sensory powers are in a much more compact package.

“Matching (or exceeding) human sensing capabilities requires AVs [autonomous vehicles] to employ a variety of sensors, which in turn requires complete sensor fusion across the system, combining all sensor inputs to form a unified view of the surrounding roadway and environment,” the report noted. “While no single sensor completely equals human sensing capabilities, some offer capabilities not possible for a human driver (e.g., accurate distance measurement with lidar, seeing through inclement weather with radar.”

A fully connected autonomous vehicle — one that employs dedicated short-range communications, for example, to exchange data with other vehicles and infrastructure — offers the best potential to effectively and safely replace the human driver when operating vehicles at automation levels 4 (high automation) and 5 (full automation), according to the study.

Brandon Schoettle, a project manager at the University of Michigan Transportation Research Institute’s Human Factors Group, authored the report.

Below is a table summarizing some of his findings:

Performance aspect	Human	AV Radar	AV LIDAR	AV Camera	Connected vehicle with DSRC	Connected autonomous vehicle
Object detection	Good	Good	Good	Fair	n/a	Good
Object classification	Good	Poor	Fair	Good	n/a	Good
Distance estimation	Fair	Good	Good	Fair	Good	Good
Edge detection	Good	Poor	Good	Good	n/a	Good
Lane tracking	Good	Poor	Poor	Good	n/a	Good
Visibility range	Good	Good	Fair	Fair	Good	Good
Poor weather performance	Fair	Good	Fair	Poor	Good	Good
Dark or low illumination performance	Poor	Good	Good	Fair	n/a	Good
Ability to communicate with other traffic and infrastructure	Poor	n/a	n/a	n/a	Good	Good

Source: “Sensor Fusion: A Comparison of Sensing Capabilities of Human Drivers and Highly Automated Vehicles,” released by Sustainable Worldwide Transportation at the University of Michigan. DSRC refers to dedicated short-range communications.

 

Follow @automotivefleet on Twitter

from Automotive Fleet http://ift.tt/2xPWWvw




Sourced by Quik DMV - CADMV fleet registration services. Renew your registration online in only 10 minutes. No DMV visits, no lines, no phone mazes, and no appointments needed. Visit Quik, Click, Pay & Print your registration from home or any local print shop.