Despite constant improvements in road safety, more and more can be done to prevent collisions between cars and other vehicles. And the problem worsens at night. Half of all fatal car accidents reportedly occur after sunset, even though there can be 75% less vehicle traffic on the roads after dark. Edge cases are a thorny issue for autonomous vehicles and explain why driverless cars are still viewed by some as a sci-fi daydream. Training self-driving cars in virtual worlds and building digital twins can help. There are other lessons too, in some unexpected places. It’s possible that nocturnal flying insects have clues on how to prevent autonomous vehicles from crashing.
Driverless cars and more
According to researchers at Penn State University in the US, insect-based collision avoidance technology could also benefit flying drones and robots. Their proposal was published in ACS Nano magazine. And an unpeered version is available as a publicly available preprint. In addition, according to the developer, the collision avoidance detector has a footprint of just a few micrometers (a few thousandths of a millimeter) square. And it hardly uses any electricity – just a few hundred picojoules.
“Currently, most collision detection techniques deployed by automakers involve the use of various sensors, including light detection and ranging (LiDAR), radio detection and ranging (radar), ultrasound and cameras to detect signals from the obstacles, followed by processing the information obtained with extensive onboard hardware or cloud servers available via internet or satellite connections,” writes the Penn State team led by Saptarshi Das. “Needless to say, these approaches incur excessive infrastructure costs and impose a huge energy burden.”
The group’s low-power design, which could potentially help prevent autonomous vehicles from crashing, is based on atomically thin and light-sensitive meme transistors (transistors with attractive storage properties). And how it works mimics the flight response of the so-called Lobula Giant Movement Detector (LGMD) neuron, found in many insect species like grasshoppers and flies. As the researchers point out, insects can outmaneuver predators, capture prey and impress mates — all while flying in the dark. But how?
Fight or flight signals
It turns out that the LGMD neuron is unique in its ability to perform nonlinear computations. And a single, task-specific neuron in each insect’s eye that “spikes,” firing electrical impulses when obstacles approach, keeps the host from bumping into things. “These spike-based computations and information processing techniques enable sensory neurobiology [insects] to perform fast, complex tasks with remarkable energy efficiency,” the researchers comment on their work. “In contrast, artificial collision detectors use continuous-time electrical signals, which increases the energy load.”
By implementing the neuron response in silicon-based electronics, the group has shown how optically sensitive components can be configured as a “spiking” night-time collision detection circuit that is highly sensitive to approaching car headlights. The setup also reacts to light reflected off walls and possibly pedestrians. Some traditional automotive collision avoidance systems can be blinded by bright light, but the insect-based approach is unlikely to suffer from the same disadvantages. Spike coding in the sensor provides a quick response that, with some development, could one day prevent autonomous vehicles from crashing.
If self-driving car developers could demonstrate that their vehicles are highly collision-avoidant, it would take the pressure off driverless car makers who are striving for fully autonomous driving. Instead of having to train AI systems for all edge cases – which is proving if not impossible but certainly a far-reaching goal – self-driving car developers could instead show authorities that their autonomous vehicle designs have a fail-safe mechanism. And learning how insects avoid collisions at night paves the way for automotive-grade accident prevention systems capable of handling riskier after-dark driving conditions.
patent submitted
Das and his colleagues have been interested in the visual information processing of insects such as grasshoppers for some time. And his group has been busy translating these abilities from the insect world into automotive solutions that could allay some of the broader concerns about self-driving vehicles. In 2020, they announced that they were in the process of patenting insect-based collision avoidance technology. Estimating a Technology Readiness Level (TRL) of 1-3, the team has shared a summary of such an ultra-low power biomimetic collision detector, including application and market benefits.
Self-driving car developers may also be able to learn from other insects beyond flies and grasshoppers to keep autonomous vehicles from crashing. The swarm intelligence of the ants could also be helpful for car navigation. In the insect world, swarm intelligence can solve complex pathfinding problems, quickly identifying the shortest path between nests and food sources. It is possible that driverless cars will use similar algorithms to coordinate themselves in traffic and draw attention to other road users.
Self-driving cars and insects, a perfect match – who would have thought?
