Scribd
Upload
25 views2 pages

ADAS & AI_ML Assessment .docx

The document outlines an assignment focused on Advanced Driver Assistance Systems (ADAS) that involves continuous data collection from various sensors like cameras and radars for improved decision-making in tasks such as object detection and collision avoidance. It details steps for understanding ADAS architecture, data preprocessing, object detection, decision-making, and improving efficiency, emphasizing the integration of AI and machine learning. The final submission should include a comprehensive documentation of the approach, techniques used, results, and visualizations.

Uploaded by

Aditya Aman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
25 views2 pages

ADAS & AI_ML Assessment .docx

The document outlines an assignment focused on Advanced Driver Assistance Systems (ADAS) that involves continuous data collection from various sensors like cameras and radars for improved decision-making in tasks such as object detection and collision avoidance. It details steps for understanding ADAS architecture, data preprocessing, object detection, decision-making, and improving efficiency, emphasizing the integration of AI and machine learning. The final submission should include a comprehensive documentation of the approach, techniques used, results, and visualizations.

Uploaded by

Aditya Aman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

ADAS & AI/ML exercise

Task
In ADAS, it is important that the vehicle gets continuous data of its surroundings accurately and
responds according to it. To achieve this, camera and radar are used. The ability to fuse data from
multiple sensors enhances the system's accuracy and robustness, providing better decision-making for
tasks like object detection, collision avoidance, and navigation assist.

This assignment aims to test your ability to work with multiple sensor inputs, perform data fusion,
and apply AI/machine learning for obstacle detection in the context of ADAS. Output can be
generated either using real life vehicle reference data or a scaled down experiment.

Steps involved

1.​ Understanding the ADAS architecture (10%)


●​ By integrating various sensors and Machine Learning models, generate a
system-level flowchart for ADAS operation.
2.​ Data Preprocessing and correlation (20%)
●​ Download and use open source dataset that includes data from different vehicle
sensors such as Camera/ Radar/ etc.
●​ Pre-process the images using transfer learning through a suitable custom model to
identify various objects (human, vehicle, bike, dog, pedestrians, etc) on the road to
generate object vectors for each of the camera images (30 camera frames per
second).
●​ Pre-process Radar data to deterministically generate the field of view and road
infrastructure. Also generate the list of objects, relative speed, and distance of each
object from the vehicle.
●​ Ensure data reliability and optimization
3.​ Object detection and Placement (25%)
●​ Correlate the object data of Camera v/s Lidar to train the AI models
●​ Perform contour based on their appearance.
●​ Identify potential obstructions dynamically in the drive path using custom AI/deep
learning models.
4.​ Decision Making (25%)
●​ To avoid collisions in least probability, visual and alarm indications should be
triggered.
●​ For confirmed collision probable scenarios, trigger corrective mechanisms like
automated braking, turn assist, lane change avoidance, ACC, etc.
5.​ Improvements to increase efficiency (20%)
●​ State at least 3 improvement points to ensure better efficiency in prediction.

Submission required Document your approach, including:


1.​ How you processed the data.
2.​ The sensor fusion techniques you applied.
3.​ How you integrated the camera and LIDAR data.
4.​ The collision avoidance algorithm and its results.
5.​ Provide visualizations of data processing, object detection, and sensor fusion results.
6.​ Provide links to source of data, models used, other reference, etc.

You might also like