Radar AI - Deep Neural Networks for Enhanced Radar Point Cloud Generation | NXP Semiconductors

デザイン・センター
トレーニング
Radar AI - Deep Neural Networks for Enhanced Radar Point Cloud Generation

Radar AI - Deep Neural Networks for Enhanced Radar Point Cloud Generation

アカウントが必要です

サインインこのコンテンツおよびサイトの追加機能にアクセスすることができます。アカウントをお持ちでないですか？今すぐご登録下さい。

About This Training

Automotive Radar is a key sensor modality to enable higher levels of autonomous driving capabilities. The state-of-the-art algorithms for the direction of arrival (DoA) estimation use analytical signal processing methods like iterative sparse code analysis and other compressive sensing techniques as well as sparse array designs to achieve higher angular resolution.

These algorithms usually have a high computational and memory cost which makes it challenging to implement them on cost-sensitive automotive embedded devices. In this training, we discuss a method that combines both analytical radar signal processing and modern deep learning techniques to create a hybrid high-resolution DoA estimation model with a low computational cost and memory footprint.

What You Will Learn

How performance radars with algorithm updates can be build
A method which combines both analytical radar signal processing and modern deep learning techniques to create a hybrid high-resolution direction of arrival (DoA) estimation model
Introduction to a highly efficient target classification neural network model for radar point clouds

Presented by

David Bethge
Product Marketing Manager Radar Systems, NXP Semiconductors
Ryan Wu
NXP Fellow, Technical Director Advanced Radar Solutions, NXP Semiconductors
Satish Ravindran
Senior Principal Engineer, Machine Learning, NXP Semiconductors

期間

55 min

言語

英語

スキルレベル

中級

0閲覧数2いいね

We are looking for ways to improve.

アンケートに答える。

Additional Resources

Driving Automation and Radar Training Academy

Premium Radar SDK

Stay In the Loop

Subscribe to our newsletter to stay updated with our latest developments and if you need further assistance, we are here to help.

Subscribe Contact Support