Introducing SAFEXPLAIN:
Safe and Explainable Critical Embedded Systems based on AI
Objectives
To improve the explainability and traceability of DL components
To provide clear safety patterns for the incremental adoption of DL software in Critical Autonomous AI-based Systems (CAIS)
To integrate the SAFEXPLAIN libraries with an industrial system-testing toolset
To create architectures of DL components with quantifiable and controllable confidence, and that have the ability to identify when predictions should not be released based on applicability’s scope or security concerns
To design, implement, or update selected representative DL software libraries according to safety patterns and safety lifecycle considerations, meeting specific performance requirements on relevant platforms
Deep Learning (DL) techniques are key for most future advanced
software functions in Critical Autonomous AI-based Systems (CAIS) in
cars, trains and satellites. Hence, those CAIS industries depend on their
ability to design, implement, qualify, and certify DL-based software
products under bounded effort/cost
Case studies
Railway: This case studies the viability of a safety architectural pattern for the completely autonomous operation of trains (Automatic Train Operation, ATO) using intelligent Deep Learning (DL)-based solutions.
Space: This case employs state-of-the-art mission autonomy and artificial intelligence technologies to enable fully autonomous operations during space missions. These technologies are developed through high safety-critical scenarios.
Automotive: This case develops advanced methods and procedures that enable self-driving cars to accurately detect road users, estimate their distance from the vehicle, and predict their trajectories while adhering to both safety and explainability requirements.
Safely docking a spacecraft to a target vehicle
The space scenario envisions a crewed spacecraft performing a docking manoeuvre to an uncooperative target (a space station or another spacecraft) on a specific docking site. The GNC system must be able to acquire the pose estimation of the docking target and of the spacecraft itself, to compute a trajectory towards the target and to send commands to the actuators to perform the docking manoeuvre. The safety goal is to dock with adequate precision and avoid crashing or damaging the assets.
Halfway through the project, RISE hosts consortium in Lund
SAFEXPLAIN consortium meets halfway through the project at RISE venue in Lund With the first 18 months of the project behind it, the SAFEXPLAIN consortium met in Lund from 16-17 April to discuss project status and next steps for the next 18 months. Great strides have...
Exploring AI-specific redundancy patterns
AI-generated image of object detection for automotive Artificial intelligence, and more specifically, Deep Learning algorithms are used for visual perception classification tasks, like camera-based object detection. For these tasks to work, they need to identify the...
Standardizing the Probabilistic Sources of Uncertainty for the sake of Safety Deep Learning
The Safexplain team from the CAOS research group of the Barcelona Supercomputing Center (BSC-CNS) presented their latest research at the AAAI's Workshop on Artificial Intelligence Safety of the AAAI 2023, held on February 14, 2023, in Washington D.C. The team's...
2023 VDA Automotive SYS Conference in Berlin
The Quality Management Centre of the German Association of the Automotive Industry (VDA) hosted the VDA Automotive SYS Conference in Berlin, Germany from 10-13 July 2023. This conference served as a platform for industry leaders and experts to discuss and showcase the...
Clustering Workshop: Establishing the next level of ‘intelligence’ and autonomy
On 2 March 2023, the SAFEXPLAIN project joined 8 other EU-funded projects under the HORIZON-CL4-2021-HUMAN-01-01 topic to share information on eachothers projects, get to know other fund recipients and explore synergies that could be pursued in common for trustworthy...