TimestampsAI

Faster and data reduced solution for high-resolution LiDAR systems

TimestampsAI is our latest development for your applications to capture complex scenes in 3D in any environment.

Distance Determination Method TimestampsAI

Our distance determination method TimestampsAI bypasses data-driven problems for all future LiDAR sensor solutions. A pixel-wise quality information given in addition to the resulting point cloud allows confident decision-making of automated systems.

With the compact feature extraction and machine learning algorithms directly based on LiDAR timestamps, future LiDAR systems can be more efficient, effective and robust.

Example system: TimestampsAI on LiDAR camera Owl
© Fraunhofer IMS

Features of TimestampsAI:

  • High data reduction rate and foreseeably less transmission and energy consumption
  • Reliable machine learning prediction
  • Short-range detection: distance determination using extremely few measurements
  • Middle-range detection: high resilience software-level under tough conditions
  • Provides pixel-wise measurement quality information

Examples of our Implementation:

 

  Short-range Middle-range
Detection range < 10 m < 60 m
Generated data before reduction 10.55 MB/s 46.6 MB/s
Data reduction rate 90 % 85 %
Measurements per prediction 10 400
Performance 0.12 m 91.15 %
Processing time 20 μs 289 μs
Background photon rate < 0.2 MHz < 5 MHz
Number of pixels 24 x 32 24 x 32

In factory automation and autonomous driving applications, complex scenes are required to be precisely captured with 3D sensors. Therein, high-resolution LiDAR systems are one promising solution. 

Such LiDAR systems generate a large amount of data, bringing great challenges to data transmission and processing on a resource-constrained embedded system.

With our services, such as TimestampsAI, you can overcome these issues and get smarter and faster LiDAR solutions for your products. Check out, what Fraunhofer IMS can do for your LiDAR systems.

 

LiDAR Data Reduction

  • Timestamp/histogram-based feature extraction
  • Design of processing workflow based on timestamps
  • Combination of digital processing and machine learning

 

Near-Sensor-Side Machine Learning Solutions

  • Timestamp/histogram-level training data generation
  • Application-specific method design using machine learning
  • Algorithm optimization on customer specified sensor systems
  • Implementation solution on embedded systems

 

Informative Point Cloud Generation

  • Point cloud simulation under various conditions
  • Point cloud generation with pixel-wise measurement quality information
  • Support for sensor fusion

LiDAR Camera Owl

192 pixel x 2 lines SPAD based camera

Publication of the Fraunhofer IMS

Data Processing Approaches on SPAD-Based d-TOF LiDAR Systems: A Review

Publication of the Fraunhofer IMS

Feature Extraction and Neural Network-based Analysis on Time-correlated LiDAR Histograms

Our technologies - Innovations for your products

Distributed Learning

Distributed learning enables training of complex tasks on multiple small embedded systems.

Hybrid Learning and PGNNs

In case of insufficient training data simulations based on physical models help to improve the data base.

Feature Extraction for Smart Sensors

By means of adapted feature extraction the size of a neural network can be reduced.

Industry 4.0 Research

Within the context of "Industry 4.0", Fraunhofer IMS is researching predictive maintenance solutions for the manufacturing industry.