Robosense develops lidar at fraction of cost

Robosense lidar system can detect objects

Robosense has developed intelligent perception lidar, a hardware and software algorithm for the mass production of safer autonomous cars. The RS-IPLS uses real-time data pre-processing and a gaze function similar to human eyes.

 

The company says the system is inexpensive, 1/400th the price of traditional 64-line lidar systems, and is designed for the mass production of vehicles at a low price.

 

Based on mems solid-state lidar, it outputs high-resolution colour point cloud data by merging the underlying hardware of 2D imagery with a deep learning sensing algorithm developed for autonomous driving. The intelligent detection algorithm achieves target level information that adjusts the region of interest (RoI) detection area in real time with no delay.

 

In early 2017, Robosense said it became the first lidar vendor to introduce laser-aerodynamic environment-aware lidar algorithms combined with lidar hardware to provide a large number of partners with its P series lidar systems based on different applications.

 

When the system’s field of view perceives a target of interest, it initiates a gaze processing mechanism that instantly locks the target for RoI processing, thus achieving clear and stable environmental data. The lidar under-the-system architecture maintains a high degree of vigilance of the surrounding environment, constantly capturing the areas of interest, allowing the gaze to transfer efficient and high-quality feedback in the field of view.

 

It also provides richer three-dimensional spatial data information (x, y, z, r, g, b) in real time from the bottom layer, which reduces the time delay normally caused by external fusion. Data pre-processing is performed by the AI algorithm, with the area of interest repeatedly detected for farther detection distance and more accurate perception results for autonomous driving, with reduced data processing stress to the central data processing unit.

 

The system is based on the firm’s solid-state lidar technology, the RS-Lidar-M1pre introduced at CES 2018. This combines the RS-LCDF lidar and camera fusion technology, lidar algorithms and gaze technology. The RS-IPLS pioneered the fusion of mems solid-state lidar technology with the Lidar-Cam deep fusion technology, AI sensing algorithm and intelligent detection technology.

 

The combination of these four technologies allows the lidar system to go beyond the sensor level, linking the sensor layer and the sensing layer at the system level, so autonomous vehicles have similar abilities as the human visual system.

 

“Lidar needs a self -revolution,” said Mark Qiu, Robosense co-founder. “The market needs a car-grade product that can be mass produced in the tens and hundreds of thousands. You have to guarantee low cost and provide high performance to ensure the safety of your vehicle. This has driven us to create a whole new sensor system technology based on the perfect extraction of environmental information and large-scale perception of the environment, with the constant capture of areas of interest, focusing on effective information.”

 

Over the past decade or so, the emergence of mechanical multi-line lidar has solved the problem of high-precision three-dimensional environment perception, and the autonomous driving industry has entered a period of growth. Lidar helps autonomous vehicles keep the promise of safety for humans, propelling the explosion of the autonomous driving industry, but at the same time, requiring more demands on lidar, such as low cost, the need for car level production, high performance and intelligence.

 

Traditional mechanical lidar is very expensive. A 64-line product currently being sold is officially priced at $80,000, more expensive than a brand new Tesla car. Robosense’s mass production price of the RS-IPLS is $200, 1/400thof the traditional 64-line lidar, with a several thousand or even tens of thousands faster response speed.

 

Traditional mechanical lidar consists of multiple sets of transmitters, receivers, precision optics, micro-control motors and other components, which are difficult to manufacture and require expensive materials. In addition, due to difficulty of integration and optics, mechanical lidar not only has a long production cycle and high production costs, but also has a difficult time meeting the stability and reliability requirements of standard vehicles. Because of these reasons, traditional mechanical lidar technology has become inadequate.

 

“The new Robosense RS-IPLS mems-based technology is intelligent, accurate, lower cost and faster, and is poised to take over the market,” said Qiu.

 

The accurate gaze function is similar to human eyes; when the field of view perceives a target of interest, the RS-IPLS will initiate a gaze processing mechanism that instantly locks the target for RoI processing to achieve clear and stable environmental data.

 

An AI-aware algorithm designed for autonomous driving improves the detection of distance, accuracy and opacity of the target object, improving the safety of the self-driving vehicle for detection with less computing power.

 

Colour point cloud with 3D dimension and RGB information technologies are fused to detect two-dimensional colour information onto 3D high-precision spatial data, providing colour point cloud data after time and space synchronisation, so it detects more details of road objects.

 

Founded in 2014, Robosense provides lidar environment-aware systems for autonomous vehicles.

www.robosense.ai

Recommend to Colleagues: 

Add new comment

Plain text

  • Allowed HTML tags: <a> <em> <p> <h1> <h2> <h3> <code> <ul> <ol> <li> <dl> <dt> <dd> <strong>
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

Follow Us:

Twitter icon
Facebook icon
RSS icon