S32Z and S32E processor families drive real-time control for time-sensitive vehicle applications
The S32Z and S32E processor families provide hard real-time, math intensive capabilities to drive advanced vehicle controls, supporting the transition from combustion to new energy vehicles (hybrid electric vehicle, plug in hybrid, full electric). The S32Z and S32E are ASIL (Automotive Safety Integrity Level) D safety capable for mission critical vehicle controls. Real-time processing applications for the S32Z and S32E include:
- Range extension
- Battery state estimation
- Electric motor control
- Advanced Driver Assistance Systems (ADAS) and autonomous safety processing
- Model predictive control
- Data-driven applications (Artificial Intelligence / Machine Learning)
Connecting real-time processors to the cloud via the S32G Service-Oriented Gateway accelerates new applications for software defined vehicles
Connecting the S32Z and S32E processors to the S32G service-oriented gateway unlocks the power of vehicle data. New features can be realized such as Intelligent data collection, data compression + edge processing to save data-transmission and storage costs, and support secure cloud connections. Over-the-air-update solutions support updates for software defined vehicle applications and the digital twin life cycle.
NXP’s collection of connected EV solutions
Over the past few years, NXP and our partners have invested in connected EV solutions to enable and accelerate intelligent, electric vehicle development. We continue to expand on this solution set as new features and applications are constantly evolving. Our current solutions include:
- Connected EV Management System
- Smart Data Access
- Automotive DevOps using Model-Based Design
- Battery Digital Twin
Connected EV Demo – publish vehicle data to the cloud
The Connected EV Management System brought together our real-time processing using the GreenBox Electrification Development Platform with the S32G GoldBox Service-oriented Gateway reference design for cloud connectivity. NXP partnered with Amazon Web Services (AWS) to enable the base platform for a cloud connected electric vehicle, whereby vehicle data (battery voltages, energy management results, trip statistics and GPS location data) are provided securely into the AWS cloud for connected mobility applications. Coupled with MathWorks® Vehicle Dynamics Blockset, we demonstrated a vehicle test track simulation with data fed in real-time to the AWS cloud.
Figure 1. Block diagram of Connected EV Solution
Smart Data Access – easy access to vehicle data
Building upon the success of the Connected EV Management System, we once again combined the GreenBox Electrification Development Platform and the GoldBox Service-oriented Gateway to create a Smart Data Access demonstration that streamlines access to vehicle data with a “no-code” approach. Collaborating with our partner aicas GmbH, we demonstrated the ability to select vehicle data and do more intelligent edge processing with the aicas EdgeSuite products. Data management features such as storage, interpolation/decimation, Java scripting, and no-code data rule definition were all possible with the aicas software solutions executing on the connected EV platform. The MathWorks vehicle simulation and AWS cloud services were also incorporated into the demonstration for an end-to-end solution.
Figure 2. Smart Data Access at the Consumer Electronics Show
Figure 3. Smart Data Access dashboard by aicas
Automotive DevOps – develop in the cloud and deploy to the vehicle edge
One of the major trends in automotive is the scaling up of development processes to support the Software-Defined Vehicle (SDV). As software development teams grow, there is a higher demand for DevOps methodology supporting Continuous Integration / Continuous Deployment (CI/CD). The solution allows large software teams to “Develop in the cloud with AWS and deploy to the edge” with NXP automotive processors. We demonstrated an Automotive DevOps workflow with AWS CodePipeline solutions – kicking off Model-Based Design builds in the cloud with MathWorks and NXP Model-Based Design Toolbox products. The model, once successfully built in the cloud, was deployed to the S32G GoldBox for embedded execution. The solution can be easily extended for models executing on the S32E GreenBox 3. The demo was supported by NXP’s High-Performance Compute Platform (“HPC”) Model-Based Development tool. The HPC supports deployment of target object files to S32G Arm® Cortex®-A53 cores, and S32Z and S32E Arm Cortex-R52 cores and the DSP/ML Processor.
Figure 4. Automotive DevOps workflow example by NXP, AWS, and the MathWorks
AWS IoT FleetWise
AWS IoT FleetWise makes it easy and cost effective for automakers to collect, transform, and transfer vehicle data to the cloud in near-real time. Automakers can select which data they want to collect from which vehicle and when they would like to collect it, minimizing transfer costs and maximizing the value of that data.
Data collected from FleetWise can be used for use cases such as:
- Train computer vision and predicative maintenance machine learning models.
- Proactively detect and mitigate fleet-wide quality issues
- Collect vehicle sensor telemetry to analyze and diagnose issues during vehicle servicing
NXP and AWS have collaborated to ensure that the S32 family of processors support the Fleetwise Edge Agent to access vehicle data on interfaces like CAN, LIN, and Ethernet.
AWS IoT FleetWise is tightly integrated with NXP’s S32G vehicle network processors and supported by the S32G Vehicle Integration Platform (GoldVIP). The S32E GreenBox 3 can send real-time data to the GoldBox for efficient, contextual data collection and secure streaming to the AWS cloud which can be used for monitoring, machine learning, digital twins, or cloud-based services. This provides a powerful capability that can give automotive customers and their partners new real-time data insights.
Figure 5. AWS FleetWise block diagram
Battery Digital Twin - improving vehicle behavior over time
A Digital Twin is the digital representation (or virtual model) of a physical object that can provide valuable operational insights. Digital twins running in the cloud can adapt over time for optimal insights and management: accurate range, accurate prediction of remaining useful life (RUL), and improved pack management. NXP’s Battery Digital Twin solution enables customers to visualize how publishing EV data to the cloud (environment, charge / discharge characteristics) can improve battery life and better predict battery end-of-life.
The Battery Digital Twin concept sends physical battery data to the NXP GreenBox 3 where a stress curve calculation is performed. Battery capacity data is sent to the NXP GoldBox which securely connects to a cloud-based virtual digital twin. The demo replays three usage profiles:
- Aggressive
- Range-optimized
- Battery Life-optimized
The digital twin is supported by a “Time Machine” where the user can move the battery age forwards and backwards in time. The Digital Twin displays the historical data based on the usage profile and projects Remaining Useful Life (RUL).
Figure 6. Battery Digital Twin demonstrator
AI/ML using NXP® eIQ Auto
Artificial Intelligence is one of the exciting applications seeing increased adoption in automotive and industrial applications. With connected solutions, customers can publish component data from the vehicle to the cloud and use the NXP® eIQ Auto deep learning toolchain to update machine learning (ML) models over time. The tools allow battery data, electric motor data, and efficiency data to funnel to the cloud in near real time. This data can train and inform neural networks to improve performance, update models for previously unseen corner cases, and then push neural network model updates back to the vehicle.
The eIQ Auto workflow is shown below:
Figure 7. eIQ Auto workflow
One more thing…
Finally, we are not stopping here. New, advanced connected solutions are being created by NXP and our partners. From Vehicle Network Digital Twin modeling to Battery Digital Twin modeling to improvements in battery cell state estimation, we continue to advance the field of Connected EV applications with leading edge processing, software, and partner ecosystem solutions.
Conclusion:
The S32Z and S32E real-time processors, combined with the S32G vehicle network processor, provide seamless vehicle connectivity. The combined solution enables current and future applications supporting deep data insights, full algorithm life cycle, and ML accelerated deployments to the vehicle edge. The high-performance computing capability of the S32Z, S32E, and S32G families along with NXP’s commitment to software, tooling, and a strong 3rd party eco-system, make us a supplier of choice for the automotive and industrial markets.