Optical Challenges in Sensor Design – A Romantic Story
By Maryse Aubé, P.Eng., M.Sc.A.
Senior Designer, Optical
Antoine de Saint-Exupéry said, “Love does not consist of gazing at each other, but in looking outward together in the same direction.”
Call me a romantic (or a nerd!), but I keep thinking that this also applies to the time-of-flight (ToF) of a LiDAR sensor. The loving couple, an emitter and a receiver in this case, are closely bundled together in a sensor.
The handsome emitter flashes in a specific direction to illuminate elements, while the beautiful receiver gazes in that very same direction and collects every little reflection coming back, noting its origin. By measuring the time lapse between the flash and its echo (the moment a reflection is collected), the position of the elements in the illuminated scene can be calculated.
Such an echo can be detected only if the reflected signal captured by the receiver rises well above ambient noise. That is not an issue when looking at a pedestrian only a few feet away, but it is definitely a bigger challenge when trying to detect a shiny black car several blocks away. Since a critical specification of a LiDAR sensor is its range (the distance beyond which an element is no longer detected), achieving higher performance requires one to increase the signal-to-noise ratio by either increasing the signal from the reflected light, reducing the noise or better yet, through a combination of both.
Increasing the Signal
In order to detect an element in the scene, the emitter must illuminate it as much as possible and the receiver must collect as many reflected rays as possible. For the receiver, this is done by maximizing the entrance aperture while tightly focusing the reflected rays on a photodetector array.
One approach is to design an emitter that manages the light so as to concentrate it on a very small segment of the scene, all the while having the receiver capture only that specific small segment of the scene. This solution provides a good range but usually requires high-speed mechanical scanning to cover the desired field of view in its entirety. The big disadvantage of this approach, however, is that the use of moving parts can adversely impact the reliability of the product.
To answer the market’s need for increased reliability, LeddarTech has developed a Solid-State approach. For instance, many of our products are based on 2D and 3D Flash technology. On the emitter side, different lasers fire up to flash in a sequence, each laser illuminating its own rectangular segment of the field of view with a laser line.
Thus, the task of the fixed receiver of the sensor is to look at the complete field of view of the sensor all at once, like a camera lens. Though static, it captures and images the reflections coming back from all those successively fired laser and focuses them as tiny spots located at different positions on a photodetector array. Each pixel of that array is in fact detecting a small section of a laser line in order to evaluate the 3D position of each specific target in the viewed scene.
The receiver section, for its part, faces several challenges:
- Maximizing the entrance aperture so that enough reflected rays from targets are captured and detected to meet range requirements, this for any angle in the entire field of view;
- Making sure light is tightly focused on the photodetector array’s pixel. This comes down to limiting any aberrations that could lead to false detections on adjacent pixels and signal reduction due to spread of detection on several pixels. Of course, again, this applies to any angle in the entire field of view;
- Limiting scattered or reflected parasitic rays that could cause false detections;
- Remaining a cost-effective solution both in terms of material-choice and manufacturing process.
A lot is asked of the receiver, but the emitter is also doing its fair share to help support and overcome challenges.
Each laser in the sensor also faces several challenges:
- Even in conditions of a wide field of view, the emitter should generate uniform laser lines to illuminate solely what the receiver is actually looking at, not one iota more. Any overflowing results in wasted optical power, which comes down to less illumination of the elements to be detected, with a corresponding decrease in the reflected signal.
- All possible means are put in place to avoid stray rays. If stray rays come to hit highly reflective elements outside the considered segments, the risk of false detections is increased.
And so, if one wants to maximize the detected signal and the range in order to obtain a highly reliable and performing sensor, both the emitter and the receiver must be closely aligned as they aim together in the same direction – like a happy marriage.
Reducing the Noise
Increasing the range of the sensor also requires reducing any noise that would consequently prevent detections. Optically speaking, this comes down to filtering out sunlight on the receiver side.
- Optical filters can either absorb or reflect sunlight, while making sure to transmit efficiently in every condition at the wavelength of the emitter;
- Detectors are selected for optimal response in the specific near-infrared wavelength range of the emitter source with low sensitivity to visible light;
- For all optical components, smart material choices help maximize transmission of the signal while allowing absorption at other wavelengths.
Hybrid Flash LiDAR – A Building Block to the Future
Leveraging strong and efficient technologies like Solid-State 2D and 3D Flash sensors in order to meet the high reliability and performance requirements of LiDAR in the automotive and mobility markets, LeddarTech’s designs are now pushing performance even further by using the Flash approach as a building block to develop a Solid-State Hybrid Flash approach. In this case, the LiDAR sensor is integrating a beam steering device, in order to segment the sensor’s whole field of view into discrete areas. As the beam steering device steers to look from one area to the next, each of these smaller areas only needs to cover a considerably smaller field of view. This type of hybrid approach can help mitigate challenges for both the emitter and the receiver, while being less demanding on the beam steering device than a conventional scanning LiDAR.
Regardless of the chosen approach to sensor design, a tight collaboration between hardware and software teams is essential to ensuring optimal results. For instance, some limitations of an optical design can be addressed through signal processing. At LeddarTech, this collaborative approach to design ensures potential challenges are quickly addressed by our leading experts from all fields who continually leverage the LCA2/LCA3 LeddarEngine and powerful LeddarSP digital signal processing software technologies to push LiDAR sensor performance even further.
Some say teamwork, I say: Happy sensor – Happy life!
P.S Curious to see how 3D flash LiDAR works? Watch this short video to see the main components in action, from the LeddarEngine to the optics configuration. Enabled by the LCA2 and LCA3 LeddarEngines, LeddarTech’s solid-state flash LiDARs provide a versatile 3D sensing solution for automotive applications (ADAS / Autonomous driving).