Smart vehicles, such as driverless cars and unmanned aerial vehicles, will need next generation radar sensors to achieve level four and level five autonomy. Smart radar sensors with much higher resolution than what is available right now are at the crux of such anticipated developments. LIDAR and cameras would have an integrally important role to play in smart vehicles. Radars are quintessential for long range detection and assessment of scenarios where sensors go blind and are hence unusable, such as inclement weathers. Existing radars would have to evolve and become much smarter to be relevant for autonomous vehicles.
The Evolution of Smart Radar
Radars are becoming smarter. There is substantial innovation underway right now. LIDARs are undergoing phenomenal changes. A new type of imaging radars has made reconstruction of environments with precision and interpretation of the world around just as humans do possible and increasingly reliable. Of course, we are a fair distance away from impeccability of such smart radar innovations. State of the art and progressively evolving radar architecture using engineered meta-material structures would be capable of beam-forming and beam-steering. Such radar architecture would be supported by artificial intelligence and hence it would be capable of detecting and recognizing, tracking and classifying a plethora of different objects in real time.
The Emergence Of The Radar Era
The second decade of the third millennium marks the beginning of the radar era. Cameras and light, detection and ranging devices have been explored sufficiently and in a myriad of circumstances in recent years. They have been transformed to meet the challenges faced by autonomous cars. Such sensors have become more reliable. There are much improved and reliable algorithms with proactive failsafe measures that ensure performance without compromising security or safety. Such transformation has commenced for radars too. The industry by and large has recognized the limitation of modern sensors and hence the utility, relevance and significance of old school radars. Traditional radars would obviously need an overhaul but that would only happen when stakeholders back such transformations.
Most automakers in the world have already realized that long range detection would be possible only with traditional radars, transformed to be compatible with modern systems and smart vehicles. This has paved the way for renewed interest in different kinds of radar technologies. Radar has clearly been an underdog amidst the evolution of smart sensors. This has partly been because of the poor resolution and resulting detection or images produced by radars. While radars operate in inclement weather, much better than sensors, they do not have any intelligence or vision. There is a large sense of assumption at play. Smart technologies can give radars eyes and perhaps ears too, there could be practical distinction of one object from another, there can be algorithm driven segregation or even real time imaging that is as conclusive as with some other high resolution sensors.
What CEOs of Top Auto Companies must Explore
Elon Musk has been emphasizing on the importance of radar. He has said time and again that radar would be the most reliable and robust sensor in automobiles. Musk has always highlighted the limitations of radar technology. Most radars can detect large objects, mostly of metal. They can place the position of an object given a mapped space and that makes way for informed assumptions. Sensors used in smart vehicles are much more nuanced and specific than that.
However, radars can be transformed to work like sensors, where they can use assisting technologies to distinguish road signs from lampposts, human traffic from automobiles and perhaps even types of cars or traffic conditions. The speed or velocity of moving objects, obstructions of any sort and much more can be effectively ascertained with the help of smart radars. Radars can become a digital eye with a much larger range than some of the best sensors available today. There are point cloud imaging solutions where rastering beams can help map objects and then an algorithm discriminates the objects rather effectively.
Artificial Intelligence And Smart Radar
AI is crucial as it would be the pivotal foundation for decentralized intelligence. Humans are the best example of decentralized intelligence. Smart cars or aerial vehicles will never be truly autonomous unless they can regulate themselves as well as humans and do away with the manmade mistakes. Right now, smart automobiles fare fairly well to avert mistakes common among humans but there are many areas where they fail miserably to even mimic the strengths of humans.
Radar sensors must have brains. There should be algorithms influencing decisions based on sensor fusion which is central intelligence and the individual sensors which form the decentralized intelligence. The combination of central and decentralized intelligences also ensures better safety and security of the vehicle, anyone inside the car and the immediate surroundings. It is not sufficient for a sensor to detect a bridge ahead. It is also necessary to know how far the bridge is so appropriate measures or maneuvers can be initiated. This will not happen with only one sensor or one type of intelligence.
Different types of sensors should generate multiple data points pertaining to the same thing so the central and decentralized intelligences can rely on the gamut of information to make an accurate assessment. It is only in such a scenario that autonomous driving will be practical given their safety standards.
Contemporary Radars for Autonomous Driving
The most advanced radars for autonomous driving available today use digital beam-forming. Analog radars such as phased array antennas are no longer in use. These are costly, consume a lot of power and are more complex than desired. They are superior in some ways but the automakers intend to keep the costs in check. This preference must change as even today analog radars are deployed to detect ballistic missiles. Autonomous driving cannot be too cost conscious. Safety and performance should be top priorities and then cost.
Limitations of Digital Radar
Digital beam-forming is slow. It needs a few milliseconds to scan a scene. The signal to interference plus noise ratio along with the processing of complex analog to digital circuitry and then assigning relevant digital weights take time and the computational load leads to substantial sluggishness. Digital beam-forming also fails to detect narrow objects, including pedestrians. Digital beam-forming also creates ghost images. It is sensitive to correlated signals and hence the noise is enhanced.
Contemporary Innovations in Radars in Automotive
Incremental engineering scaling up from tier one to tier two and forward is aimed at improving MMIC technology. Startups are exploring uncharted territories such as Luneburg lens antennas, algorithms that can suppress interference in digital beam-forming and three-dimensional printed arrays. Not all new creations will survive the test of time. Even if there is a few viable developments, that would change the scene for radars.
Challenges of Implementing a New Type of Radar
The frequency range available is 76 GHz to 81 GHz. There is a shortage of materials and components. Many components available are not characterized. Forensic work is extensive and hence a challenge. There are other challenges that are more manageable and would be overcome in due course of time.
Induction of Smart Radars in Autonomous Vehicles
Smart vehicles are expected to achieve level five autonomy by 2025. By the end of this decade or as soon as next year, smart radars would witness a quantum leap in technology, utility, reliability and safety. This shall be followed with reasonable deployment.
The Distinct Advantage of Metawave Smart Radar
The use of artificial intelligence and adaptive meta-materials remain the strengths of Metawave. Smart radars capable of working at fascinating speed and accuracy, boasting of possession superior vision and being driven by reliable practical intelligence are poised to be game changers.