Cheap lidar: the key to making self-driving cars affordable
Have you been in a position to see a self-driving vehicle out there on the streets? If you have, you must have noted the spinning sensor on the roof of these vehicles. The sensor is what helps the autonomous car to understand its immediate surroundings. However, the sensor and the technology supporting it is what are making the self-driving cars unaffordable to the average consumer.
Light radar commonly referred to as Lidar is quite expensive owing to its complexity. However, it is shortly envisioned that the Lidar technology will be more reliable, cheaper and about everywhere.
Lidar just like radar is a sensing technology that will detect objects using pulses of laser light. The difference lies in the fact that lidar has shorter range due to the shorter wavelengths resulting in greater resolution. Lidar, therefore, produces high quality and reliable data. Lidar is, therefore, the sensor of choice for the autonomous vehicle applications.
Experts are unanimous that lidar is a necessity for the driverless vehicles. Karl Iagnemma, CEO at nuTonomy that is currently testing the autonomous vehicles in Singapore, says that lidar is a necessity because it is very precise and has robust application in a wide array of conditions. He also notes that the complexity, size, and cost of lidar sensors are presenting important challenges in the commercialization of applications dependent on the lidar technology.
The commonly used HDL-64E lidar sensor from Silicon Valley–based Velodyne weighs a massive 13 kilograms and will set you back $80,000. Despite the shortcomings, the sensor can scan 2.2 million data points within its field of view per second and can identify the location of an object up to 120 meters away with centimeter accuracy.
Velodyne has announced the VLP-32A sensor which weighs 600 grams, has a 200-meter range and will cost $500. It is an improvement to the earlier generation sensor but still too expensive for autonomous car production aimed for the consumer market.
Substantial industry and academic research are ongoing to try and make the lidar sensor smaller and cheaper. Quanergy Systems demonstrated a prototype solid-state lidar sensor at the CES 2016 electronic show. The sensor will make use of the optical phased array to steer laser pulses as opposed to the rotating system of mirrors and lenses. Quanergy explained that its sensor will cost $250 in volume production and will be available to automotive equipment manufacturers in early 2017.
Two other startups, Innoluce and Innoviz, are working on $100 automotive lidar systems with their release planned for 2018. Innoviz from Israel promises a high definition solid-state lidar with a larger field of view and greater resolution. Innoluce from Netherlands has taken a different approach with the use of microelectromechanical mirror system to scan and steer a laser beam.
The most exciting and promising development is happening at MIT with funding from the U.S. Defense Advanced Research Projects Agency’s Electronic-Photonic Heterogeneous Integration program. The researchers at MIT have come up with a 0.5- by 6-millimeter chip by leveraging silicon photonics to condense a functional lidar system that can be fabricated in commercial CMOS foundries. The only challenge is that the MIT prototype has a range of a few meters. However, efforts are ongoing to get to the 100-meter range and retain a per-chip cost of $10.
A flurry of activity in the small and cost effective solid state lidar system cannot go unnoticed. In August, Quanergy obtained a $90 million funding round on a $1.59 billion valuation whereas Ford and Chinese search giant Baidu jointly invested $150 million in Velodyne. The efforts are aimed at making a 100 automotive lidar sensor available within the next couple of years.
Even though the automotive industry is the largest market for low-cost lidar at the moment, it is without a doubt that small and inexpensive lidar sensors will benefit many other industries. Other applications of lidar technology are in robots enabling recognition and manipulation of objects with greater confidence, in mobile phones to allow high-resolution gesture recognition, and in drones to help them avoid obstacles making them safer for tasks like package delivery.