Occasionally, every few months, there's some news about Apple's car. Two months have passed, and it seems like the only major news from Apple has been the iPhone X.
Fortunately, Apple's progress in the automotive sector hasn't gone unnoticed entirely. According to Reuters, two Apple researchers who are working on autonomous driving technology recently published a paper online discussing how self-driving cars can detect cyclists and pedestrians using fewer sensors.
The perception, decision-making, and control of driverless vehicles are crucial steps in simulating human driving behavior. Among these, perception is the most fundamental and decisive factor that influences the final control outcomes.
On-board computers have limited processing power. If too many sensors are used, more data is collected, but this also increases the computational load. With current on-board computing capabilities, processing large amounts of data takes longer, which can delay vehicle control decisions.
Therefore, for each autonomous driving research team, simplifying the sensor setup while still collecting enough information is key to improving the overall efficiency of the autonomous driving algorithm.
The paper submitted by Apple was posted on the open-access journal arXiv on November 17th, marking the first time Apple publicly disclosed its self-driving technology. The core innovation in the paper is how Apple’s new software improved the LiDAR system’s ability to identify pedestrians and cyclists from a distance.
This can be considered a significant breakthrough in Apple’s autonomous vehicle technology. A while back, images of Apple’s driverless test car showed a very complicated and outdated sensor setup. If Apple can implement the solution described in the paper, its self-driving vehicles may have a better chance of competing in the autonomous industry.
Currently, most self-driving systems use a hybrid sensor approach, combining LiDAR, cameras, and millimeter-wave radar to achieve full autonomy. Each sensor has its strengths and weaknesses, so they’re used together to complement each other.
LiDAR works by emitting light beams to measure the distance and shape of objects. Compared to image-based detection, LiDAR provides depth information, allowing for more accurate object positioning and identification. However, due to factors like uneven 3D sampling and range limitations, LiDAR point clouds can be sparse and less reliable.
Apple’s team claims they’ve largely solved this issue. They say their software can reliably detect cyclists and pedestrians using only LiDAR data and can distinguish them from other 3D objects. However, testing is currently limited to simulations, and no real-world road tests have been conducted yet.
Despite this, Apple has adjusted its self-driving car to rely heavily on LiDAR as the main sensor. A month ago, MacCallister Higgins, founder of startup Voyage, released a video showing Apple’s third-generation self-driving car.
The video reveals a white bracket mounted on the roof of a Lexus RX450h. Each bracket contains three cross-shaped sensor arrays: vertically, two Velodyne 16-line or 32-line LiDAR units, and horizontally, two millimeter-wave radars (the exact models aren’t clear). In the center of the cross, there’s a camera.
These six sensor arrays cover the left front, front right, left rear, front, and right rear directions, providing a 360-degree view around the vehicle.
It's hard to say whether this sensor setup is cost-effective, as LiDAR is expensive and difficult to conceal, making the car look more like a "test vehicle" than a regular car.
In contrast to other companies aiming to bring affordable autonomous tech to market, Apple's focus on high-end LiDAR solutions suggests a more long-term strategy.
While others are racing to deploy low-cost sensor systems, Apple is still refining LiDAR performance. This slow pace might not be ideal, especially in a competitive field like autonomous driving.
Apple's closed approach in consumer electronics has led to groundbreaking products, but in areas like AI and autonomous driving, it doesn’t seem to attract as much talent. Slow progress could lead to losing key personnel over time.
Recently, a Silicon Valley startup named Aeva launched a new sensor with high-precision distance measurement and speed detection—essentially meeting the needs of autonomous vehicles. Its founders previously worked on Apple’s self-driving program.
Autonomous driving is now a highly competitive space, with many players racing to commercialize fully autonomous vehicles by 2021. Although Apple started late, it has already released a completely new version of its self-driving system in less than a year, showing strong determination.
Even though many autopilot startups are seeking funding and trying to launch businesses, the reality is that autonomous driving is a long-term, money-intensive endeavor. Investors won’t stay patient forever, and eventually, many talents will be absorbed into larger companies.
The competition is just beginning, but the future will soon become clearer. We're looking forward to seeing what Apple will present next in its autonomous driving research.
Remote Controller Socket Converters
Remote Power Socket,Remote Control Socket Adaptors,Remote Control Plug Adapters,Remote Control Socket,Remote Power Switch
NINGBO COWELL ELECTRONICS & TECHNOLOGY CO., LTD , https://www.cowellsockets.com