Government Funding News
» » » » Connected and Autonomous Vehicle Technology R&D Trends

Connected and Autonomous Vehicle Technology R&D Trends

While the transition to full autonomy will be gradual, technology developers and manufacturers need to be well ahead of consumer demand to push the technological readiness of autonomous vehicle systems. Constant research, development, and testing will help ensure that autonomous and connected vehicle technologies can be deployed once market and regulatory conditions support implementation. Having ready, proven technologies can help accelerate this transition by proving to consumers and government regulators that autonomous vehicles are safe and capable of improving the driving experience.

Considerable opportunities exist for technology developers that become early leaders in the autonomous vehicle revolution.

Connected and autonomous vehicle systems rely on many components to operate safely and deliver value to consumers. To remove humans from vehicle operation, autonomous vehicles must deploy a range of sophisticated technologies such as cameras and sensors, lidars and radars, mapping and GPS, software and machine learning, and smart infrastructure.

Autonomous Vehicle Cameras and Sensors

Autonomous vehicles deploy many technologies to see and understand the world around them. Although cameras and sensors pre-date autonomous vehicles, they play a vital role in C/AV success by inputting data based on how close they are to other vehicles or objects. Both are important but serve unique functions. Cameras are primarily used to interpret visual cues that are two-dimensional or outside of the vehicle’s movement path, including signs and lines on the road. Ultrasonic sensors are less intelligent, but just as useful. Sensors help give autonomous vehicles better depth perception, but generally cannot recognize what certain objects are.

Camera and sensor technology developers have significant opportunity to develop components that can see further in extreme conditions and better recognize obstructions.

Two central issues related to the technological readiness of cameras and sensors focus on reliability. Although cameras can provide vehicles with detailed information about surroundings and can be better used to determine the nature of an obstacle, they are less reliable in adverse weather conditions like heavy rain or snow. Ice build-up is also an issue, since camera lenses covered in ice cannot see and may be difficult to clear off.

Sensors, on the other hand, are less vulnerable to poor weather conditions since their ultrasonic signals can penetrate rain and snow to identify obstructions. Although ice build-up can still affect performance, they are more reliable. Sensor technologies can be further developed to identify the size and shape of obstructions to help understand what they are. Instead of a simple “obstruction or no obstruction” determination, the sensor should be able to propose how close the obstruction is and deliver cues to the vehicle’s software to help it determine whether the vehicle needs to stop or change its course of direction.

Autonomous Vehicle Lidars and Radars

Other, more revolutionary equipment that can help drive autonomous vehicles are lidars and radars. Although radars have been included in vehicle production for the last two decades, they’ve never been used to the degree that autonomous vehicles will require. It’s likely that, given their performance in adverse conditions, radars will be standard on all early-stage autonomous vehicles. Beyond the first generation of driverless vehicles, lidars may play more of a role because it uses more sophisticated technology that can also deliver better data to the vehicle’s processing software.

Radar data can be improved to offer better analysis of a vehicle’s environment, while lidar technologies should be innovated to offer better reliability at less cost.

Although you can’t see them, radar technologies have helped vehicles identify and alert drivers to objects on the road for about 20 years. Radar technologies are often located under a vehicle’s sheet metal and can assist with driving functions that assess object speed, such as adaptive cruise control and automatic emergency braking. Unfortunately, radars still cannot “see” to the extent of a human driver, and currently do not provide much detail to a vehicle’s operating system. While radar technology will have use in autonomous vehicles, it will need to work more closely with other systems to assemble a true, real-time understanding of the vehicle’s environment.

Eventually, radar systems will become obsolete because of improving lidar technology. Current lidar systems use spinning components to shoot millions of light pulses in every direction and measure how long they take to bounce off other objects. Once light signals return to the lidar, it can build a rapidly-updated map that establishes the position (and even velocity) of surrounding objects. What’s preventing these systems from being installed on current autonomous vehicles is overall cost while striking a balance between signal range and resolution. Spinning systems are also less likely to be effective in Canadian winters, so developing a reliable, stationary system will help roll-out the technology to more consumers.

Mapping and GPS for Autonomous Vehicles

All four of the technologies discussed so far were individual components that identify a vehicle’s surroundings. Data generated by these systems needs to be pieced together in a way that the vehicle’s operating system can understand, which is what mapping technologies do. Alternately, GPS systems offer a location and global path for a vehicle to follow. GPS can suggest multiple routes to transport passengers to a destination but does not maintain the intelligence to influence safe commuting without other systems.

Autonomous vehicles must merge these two systems, using GPS for route navigation while also using mapping to apply the right vehicle functions at the right time.

Most current autonomous vehicles use Bayesian simultaneous localization and mapping (SLAM) algorithms to combine data from sensor systems and an offline map. Together, this data shows current location estimates and map updates.

Advanced mapping systems may leverage other data to build a bigger picture of vehicle navigation, including tracking of other moving objects such as vehicles and pedestrians. This way, instead of working from an offline map, the vehicle continuously updates a map of its surroundings and factors them into route efficiency. Mapping could also be shared between vehicles to build a global database rather than relying on simplistic maps with single-vehicle data input.

C/AV Software and Machine Learning

Software is responsible for processing data inputs and instructing autonomous vehicles to perform an action. While this may seem simple in a basic stop/go scenario, real driving requires that C/AV software be incredibly complex. Software must not only see, but truly understand its environment to make correct determinations. It must recognize and differentiate between obstructions to make critical navigation decisions. It should also be capable of becoming more intelligent as it’s used more and has additional data on common routes or driving conditions.

Technology developers have significant potential to develop autonomous vehicle software that avoids collisions and ensures pedestrian/cyclist safety.

Safety is a critical issue preventing widespread adoption of autonomous vehicles. While testing shows that, even in mixed traffic, autonomous vehicles are safer than human-controlled vehicles, every instance of autonomous vehicles crashing or endangering humans is a setback to market acceptance. It may be impossible to reach a point where accidents never occur, but people need to feel that it’s safe to be around autonomous vehicles. Intelligent software that constantly learns from its experiences is a key to reducing accidents and calming fears from the public or government regulators.

Connected and Autonomous Vehicle Smart Infrastructure

One must also consider the critical role infrastructure or grid technologies will play in the future of autonomous driving. Many types of technologies will be rolled-into traffic lights, street lights, and signs, and may even be built into roadways or other permanent structures. By developing external systems that enable autonomous driving, less emphasis is required on building so much technology directly into vehicles. Vehicles will share infrastructure sensors and signals, helping navigation along driving routes.

Development and maintenance of shared, public infrastructure will create significant opportunity for technology developers.

While it’s still unclear how big of a role vehicle-to-infrastructure technologies will play, it seems that many large cities have begun adopting and testing such equipment. Deployment is difficult because there’s no way to enforce a standard solution across the globe, but there will be islands of opportunity here where more progressive municipalities or regions invest in smart infrastructure. Ultimately, it seems infrastructure technologies will be more important in situations where vehicle-to-vehicle communication is limited, either because of weather, geography, or other factors.

Next: Access Industry Resources & Canadian Government Funding

Is your company capable of innovating any of the connected and autonomous vehicle technologies listed in this article? If so, there’s a wide range of resources including mentorship and government funding programs that can help turn your innovative concept into a working product.

To learn more about these resources, please download our Electric and Autonomous Vehicle Trends white paper.

X