Nerd For Tech
Published in

Nerd For Tech

Lidar on the Car, the Next Step in Autonomous Driving?

From 2021 to 2030, autonomous driving has entered a golden decade of rapid development.

Public information shows that by 2030, 20%-25% of vehicles in the market will have high-level driving assistant functions, similar to Tesla’s NOA.

It means that there will be a standard space of 4–5 million vehicles per year, and the market potential is beyond imagination.

Mastering Data Fusion Can Master the Future

“During the entire golden decade of autonomous driving, we can regard the development of automobile intelligence as an exponential function, first moving forward smoothly, then accelerating by leaps and bounds, soaring into the sky. Therefore, the first five years can be regarded as a key turning point for the industry. Whether it is OEMs, lidar manufacturers, or other R&D manufacturers, they must seize this growth period.

Along with this rally, there is also a rigid demand for high-level data fusion in the industry, and it is increasing. Among them, the route of using lidar as a sensor and fusing other perception data has been verified and deployed on a large scale in L4 scenarios such as Robo -taxi and low-speed logistics. However, there are still many challenges in applying high-level autonomous driving to passenger cars.

“At present, the algorithms related to target perception and target decision-making based on the laser point cloud itself are not enough in the industry, and the accuracy and training level of the model is not very high.” Li Dongmin said.

Secondly, lidar should also perform corresponding fusion calculations with other sensors, high-precision maps, and inertial navigation.

It is a new topic and proposition for the entire automotive industry. In terms of the productization of technology, the entire industry still has a very large technological gap to be bridged.

To cooperate with OEMs to overcome the problem of mass production of passenger vehicles, a technology company has previously carried out a series of practices of laser point cloud-based fusion perception capabilities and developed a number of point cloud algorithms. It has formed a mass-produced fusion perception solution to achieve the fusion perception of multi-channel vision, including environmental perception, situational awareness, and behavior prediction.

Fusion Perception Is A Key Step in Unmanned Driving, and It Is Also the Most Researched Topic at Present.

The company will use the fusion perception algorithm to fuse the raw data of vision, point cloud, and millimeter-wave for calculation, and then perform 3D tracking to output more accurate and complete perception results. At the same time, with the help of high-precision maps and road-end global samples, the behavior trajectories of traffic participants can be accurately predicted.

“However, the biggest technical difficulty faced by multi-sensor fusion is to achieve the synchronization of time and space.” An engineer introduced.

Much practical experience shows that whether it is for open, semi-open, or closed roads, whether it is commercial vehicles or passenger vehicles, the requirements for full-stack fusion solution providers are very high if you want to horizontally open different scenarios, which is equivalent to meeting the custom requirements of each business scenario.

The Industrialization Focus of Autonomous Driving

The mainstream algorithm model of autonomous driving is mainly based on supervised deep learning. It is an algorithm model that derives the functional relationship between known variables and dependent variables. A large amount of structured labeled data is required to train and tune the model.

On this basis, if you want to make self-driving cars more “intelligent”, and form a closed loop of the business model for self-driving applications that can be replicated in different vertical landing scenarios, the model needs to be supported by massive and high-quality real road data.

2D-3D Fusion Data:

For example, in order to develop multi-model machine learning algorithms for self-driving cars, some manufacturers need to fuse two distinct data sets with different dimensions. This operation is essential, but it is challenging to perform manually.

AI companies even hope that data companies can better understand algorithm technology and demand scenarios, participate in the research and development of algorithms, and give optimization suggestions on data collection. It has become the focus of data service providers to create a competitive advantage as well.

Common Data Labeling Types Include:


Outsource your data labeling tasks to ByteBridge, you can get the high-quality ML training datasets cheaper and faster!

  • Free Trial Without Credit Card: you can get your sample result in a fast turnaround, check the output, and give feedback directly to our project manager.
  • 100% Human Validated
  • Transparent & Standard Pricing: clear pricing is available(labor cost included)

Why not have a try?




Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store



Data labeling outsourced service: get your ML training datasets cheaper and faster!—