Technical Challenge:
Design and Implement an Explainable AI System for Real-time Anomaly Detection in Multimodal Sensor Data from Autonomous Vehicles.
Background:
In the context of autonomous vehicles, safety and reliability are paramount. Advanced driver-assistance systems (ADAS) and full autonomous vehicles rely on sensors to detect and respond to various stimuli. However, these systems are prone to errors due to the high dimensionality and complexity of the data generated by multiple sensors (e.g., cameras, radar, lidar, and ultrasonic sensors).
Challenge:
Develop an Explainable AI (XAI) system that can detect anomalies in real-time multimodal sensor data from autonomous vehicles. The system must:
- Handle Multimodal Sensor Data: The system should be …
Technical Challenge:
Design and Implement an Explainable AI System for Real-time Anomaly Detection in Multimodal Sensor Data from Autonomous Vehicles.
Background:
In the context of autonomous vehicles, safety and reliability are paramount. Advanced driver-assistance systems (ADAS) and full autonomous vehicles rely on sensors to detect and respond to various stimuli. However, these systems are prone to errors due to the high dimensionality and complexity of the data generated by multiple sensors (e.g., cameras, radar, lidar, and ultrasonic sensors).
Challenge:
Develop an Explainable AI (XAI) system that can detect anomalies in real-time multimodal sensor data from autonomous vehicles. The system must:
- Handle Multimodal Sensor Data: The system should be able to process and analyze data from multiple sensors with different data formats and sampling rates.
- Detect Anomalies: Identify anomalies in real-time, such as unusual sensor readings or patterns, that may indicate system failure or external interference.
- Provide Explainability: Offer insights into the decision-making process, including the relevant sensor data and features used to make the anomaly detection.
- Meet Runtime Constraints: The system should be able to analyze data at a rate of at least 10 Hz (every 100 milliseconds) to ensure real-time detection.
- Meet Memory Constraints: The system should be able to operate within a memory budget of 8 GB of RAM and 50 GB of storage.
Evaluation Criteria:
- Anomaly Detection Accuracy: Evaluate the system’s ability to detect anomalies correctly (True Positive Rate) and incorrectly (False Positive Rate).
- Explainability: Assess the system’s ability to provide clear and actionable insights into the decision-making process.
- Runtime Performance: Measure the system’s ability to analyze data at a rate of at least 10 Hz.
- Resource Utilization: Monitor the system’s memory and storage usage.
Submission Guidelines:
- Participants must submit a comprehensive report detailing their system design, implementation, and evaluation results.
- Code and data used for evaluation must be made available upon request.
- Submissions will be evaluated based on the evaluation criteria above.
Prizes:
- Best Anomaly Detection Accuracy: $10,000
- Best Explainability: $8,000
- Best Runtime Performance: $8,000
- Best Resource Utilization: $6,000
Deadline: February 28, 2026
Contact: For more information and submission guidelines, please contact dr.carlos.ruizviquez@ieee.org
Publicado automáticamente con IA/ML.