Multi-SoC Fail-Safe Orchestration for Autonomous Driving Workloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automotive systems rely on expensive, high-quality silicon for fail-safe operations, which are limited to handling non-catastrophic partial SoC failures and lack scalability, as they are based on single SoC solutions rather than clusters.
Innovation Solution
A fail-safe system utilizing a local compute cluster with multiple SoCs and software orchestration to prioritize and migrate critical applications, ensuring continued operation even if a node fails, by de-prioritizing lower priority tasks and redistributing workload across remaining nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional single SoC solutions with automotive grade silicon are used for fail-safe operations, then reliability is improved, but cost increases and scalability is limited
Solution Approach 1:
The patent uses software virtualization to create virtual copies of critical applications that can run on consumer-grade silicon. Instead of requiring expensive automotive-grade silicon for redundancy, the system creates software-based copies of safety-critical functions that can be migrated between physical nodes, effectively copying the functional behavior rather than requiring physical hardware duplication with high-quality silicon.
Solution Approach 2:
The patent replaces expensive automotive-grade silicon with cheaper consumer-grade silicon for the compute nodes. The system accepts that individual nodes may fail (short-living in terms of operational reliability) but uses rapid software-based failover to maintain overall system reliability, treating individual hardware nodes as disposable rather than requiring permanently reliable expensive silicon.
2Device complexity
If single SoC solutions are used for fail-safe operations, then implementation is simplified, but scalability is reduced
Solution Approach 1:
The patent segments the fail-safe system into multiple independent SoC nodes that can be distributed across the vehicle. Each node runs virtualized critical applications, and the system architecture divides responsibilities between compute nodes, network infrastructure, and software orchestration layers. This segmentation enables scalable deployment from small to large vehicle fleets while maintaining fail-safe capabilities.
Solution Approach 2:
The patent creates a universal software platform that can run on multiple types of hardware nodes. The virtualization framework and orchestration system are designed to be hardware-agnostic, allowing the same fail-safe software architecture to operate across different SoC configurations, vehicle types, and deployment scenarios, thereby achieving both simplicity and scalability.
3Ease of manufacture
If consumer-grade silicon is used instead of automotive-grade silicon, then cost is reduced, but reliability under harsh conditions deteriorates
Solution Approach 1:
The patent introduces a software intermediary layer (virtualization platform and orchestration system) between the consumer-grade silicon hardware and the safety-critical applications. This software layer abstracts the hardware limitations and provides reliable execution environments, compensating for the lower inherent reliability of consumer-grade silicon through software-based fault isolation, monitoring, and rapid failover capabilities.
Solution Approach 2:
The patent implements preemptive software-based protection mechanisms including virtualized execution environments, real-time monitoring of application health, and pre-configured failover paths. These cushioning measures are established before failures occur, allowing the system to maintain reliability under harsh conditions by detecting and responding to hardware limitations before they compromise safety-critical operations.
Data Source
AI summary
Apparatuses, methods and storage medium associated with embedded computing, are disclosed herein. In embodiments, an embedded computing platform includes a plurality of system-on-chips (SoCs) forming a local compute cluster; and an orchestrator disposed on one of the SoCs arranged to orchestrate fail-safe operations, in response to a reported unrecoverable failure requiring shut down or partial disabling of one of the SoCs, to consolidate execution of critical workloads on one or more of remaining fully or partially operational ones of the SoCs. Other embodiments are also described and claimed.