Novelic’s ACAM system is a cutting-edge automotive radar technology designed to monitor the vehicle cabin and detect the presence and movement of occupants, including children. It enables reliable in-cabin sensing capabilities that help automotive manufacturers meet emerging safety standards such as Euro NCAP’s child presence detection (CPD) protocols, which aim to prevent tragic incidents involving children left unattended in vehicles.

The ACAM solution from Novelic has received industry recognition for its innovative approach to in-cabin sensing, including the In-Cabin Innovation Award.

ACAM provides industry-leading CPD performance in a compact, power-efficient module, while also offering additional functionality, such as seat occupancy detection, and intrusion and proximity alerts. During the development of ACAM, Novelic used the HighTec Development Platform, including HighTec’s automotive-grade C/C++ compiler optimized for the Infineon Aurix TC3x microcontroller architecture, to develop and optimize the embedded software powering its radar-based monitoring capabilities.

The collaboration between Novelic and HighTec also enabled faster development cycles and reduced time-to-market. HighTec’s experienced technical support team worked closely with Novelic engineers throughout the development process, helping to resolve technical challenges and ensuring that development milestones could be achieved on schedule.

HighTec’s multi-architecture compiler support gives developers the flexibility to support multiple processor platforms and future hardware generations, allowing companies like Novelic to maintain long-term platform flexibility as automotive electronic architectures continue to evolve. Because HighTec compilers are based on widely adopted open-source technologies, development teams benefit from a shorter learning curve and reduced vendor lock-in, enabling engineers familiar with open-source toolchains to become productive quickly and integrate the tools easily into existing development environments.

HighTec provides native support for modern C++ standards, enabling engineering teams to use current language features to build scalable and maintainable software for future vehicle platforms. Its compilers are designed for demanding applications such as in-cabin radar monitoring, where systems must process complex signals and detect subtle movements, like breathing, in real time under strict power and performance constraints. The compilers are optimized for high-performance embedded software and are certified to the ISO 26262 functional safety standard up to ASIL-D, the highest automotive safety integrity level, providing a reliable foundation for safety-critical development.

Dr Raffaele Soloperto, VP for radar solutions at Novelic, said, “The development of ACAM required a robust and reliable software development environment capable of supporting advanced radar signal processing while meeting automotive reliability standards. HighTec’s Development Platform enabled our engineering team to efficiently build and optimize the software powering our ACAM solution.”

“Novelic’s ACAM solution demonstrates how advanced sensing technologies are transforming vehicle safety systems,” added Mario Cupelli, CTO at HighTec. “We are proud that the HighTec Development Platform supported the software development behind this award-winning radar-based in-cabin monitoring solution and contributed to enabling safer vehicles for the future.”

In related news, study finds Mazda ADAS bundles deliver compounding crash claim benefits

The analysis covered Mazda vehicles from model years 2015 to 2023, examining six ADAS bundles and four standalone systems. Technologies assessed included automatic emergency braking, lane departure prevention, high beam assist and driver monitoring systems.

According to HLDI, larger bundles of safety features, typically incorporating newer-generation systems, were associated with greater reductions in claim frequency.

The most basic bundle, which includes front AEB with forward collision warning, reduced property damage liability (PDL) claims by 13% and bodily injury liability (BIL) claims by 9%. The most comprehensive bundle, adding features such as pedestrian detection, adaptive cruise control, rear AEB and driver attention alert, reduced PDL claims by 39% and BIL claims by 21%, although the latter was not statistically significant.

Two systems, front AEB with pedestrian detection and rear AEB, delivered the most notable additional reductions: updated AEB systems improved prevention of vehicle-to-vehicle crashes, while rear AEB was particularly effective in reducing low-speed parking collisions.

Standalone systems also showed a measurable impact. Blind spot monitoring combined with rear cross-traffic alert reduced PDL claim frequency by nearly 10% and BIL claims by 13%. Other features, including curve-adaptive headlights and head-up displays, were linked to smaller reductions. Traffic sign recognition did not show clear benefits in this dataset, which HLDI attributed to system limitations or lower adoption rates.

While some ADAS features were associated with increased claim severity due to the cost of replacing sensors, overall losses, combining claim frequency and severity, were generally lower for vehicles equipped with Mazda’s safety systems.

“Another important factor is that crash avoidance systems primarily eliminate crashes that occur at slower speeds,” said Matt Moore, chief insurance operations officer at HLDI and the Insurance Institute for Highway Safety. “That takes low-dollar claims out of the equation and skews the average cost upward.”

Mazda said the findings support its approach to integrating multiple safety technologies across its vehicle line-up.

“As this independent analysis demonstrates, continual improvement of driver assistance technologies has real world impact,” said Jennifer Morrison, director of vehicle safety strategy at Mazda North American Operations. “We remain committed to advancing both the performance and availability of these systems in our pursuit of zero fatal crashes.”

The study highlights the role of bundled and continuously updated ADAS features in improving real-world safety outcomes.

Related news, Mercedes-AMG leverages Track Sport Concept for next-gen GT3 and Black Series

Wheel position and orientation measurements are critical for applications including wheel arch and component clearance verification, elastokinematic simulation validation and chassis and tire development. Stable and reproducible data is especially important in motorsport and vehicle dynamics control systems such as ESC, torque vectoring and advanced driver assistance systems and automated driving, particularly under challenging conditions including dust, water spray and high vibration.

Efficient measurements

The heart of the sensor, consisting of several encoders and electronics, has been completely redesigned, with magnetic encoders replacing the previous optical encoders.

The RV-5 connects directly to the wheel carrier structure and measures the wheel vectors at the source. Consequently, measurements remain stable and reproducible in rain, dust, splash water or abrupt changes in light. The factory calibration in the vehicle coordinate system remains stable across temperature and vibration changes; recalibrations, as required with camera-based setups, are not necessary.

The sensor resists optical interference and delivers reliable, high-precision, low-latency data on rough roads, curbs and under high lateral and vertical accelerations, even where camera- and optics-based systems struggle.

Magnetic encoder

The magnetic encoders provide higher linearity and stability, delivering clean, high-resolution data across the full range, even under dynamic loads. Lower sensor noise produces smoother raw signals that require less filtering and are available with reduced latency.

Oliver Plenter, business development manager at Kistler, said, “Magnetic encoders make our wheel vector sensors even more robust and precise without disrupting customers’ established setups. Plug in, configure, measure – everything as usual, but with even higher quality.”

Related news, BeyondMath raises US$18.5m to scale foundational physics AI model

The automotive industry is moving towards the supply of passenger cars with a higher-voltage electrical system, a 48V system, rather than the traditional 12 V systems, offering thinner power supply cables and reducing wire harness weight. However, in-vehicle networks must handle higher-voltage shorts, and loss of ground (GND) is a concern. As a result, PMA sublayer components – such as PHYs, system base chips (SBCs) and standalone transceivers – must be designed to survive these conditions.

The CiA specifications will extend the existing requirements given in ISO 11898-2 (CAN) and ISO 17987-4 (LIN), and CiA will specify related test cases for 48V shorts and loss of GND.

The SIG is chaired by Marko Moch, working with Cariad, a VW Group member.

Related news, Emerson expands NI PXI platform with lower-cost, high-performance test hardware

The integration aims to accelerate lidar-based perception by streamlining the process of annotating and labeling sensor data for AI model development. Vueron’s VueX platform leverages advanced AI to automatically annotate lidar point clouds with bounding boxes and object classifications, reducing the manual effort required while maintaining the precision needed for safety-critical applications. Users can upload Innoviz lidar data directly to the cloud-based VueX platform, where automated annotation tools process the data and allow for fine-tuning to achieve optimal accuracy.

Building on an existing partnership, Vueron has expanded its autonomous driving platform to support Innoviz’s latest lidar solutions. Following the successful use of the first-generation InnovizOne, the VueX platform now supports InnovizTwo for automotive applications and InnovizSmart for smart infrastructure, enabling an end-to-end AI cloud environment for data storage, visualization, processing, labeling, training, validation and deployment.

“Our collaboration with Vueron represents an important step in making advanced lidar technology more accessible and easier to deploy,” said Omer Keilaf, CEO and co-founder of Innoviz. “By integrating our automotive-grade sensors with Vueron’s powerful AI development tools, we’re enabling our customers to accelerate their development timelines and potentially reduce the complexity of bringing perception systems to market.”

“Integrating Innoviz’s high-performance lidar data into our VueX platform allows us to offer customers a comprehensive AI-based solution across automotive and infrastructure applications,” added Noah Jang, SVP of global business at Vueron. “The combination of Innoviz’s proven sensor technology with our automated annotation and training capabilities creates a powerful toolset that addresses real pain points in perception development.”

In related news, Ambarella unveils DevZone for rapid edge AI development

The original MPS was launched just over a decade ago and ,according to the company, was the first of its kind with its compact, self-contained design and built-in web server. According to Evolution Measurement, it was the world’s first miniature pressure scanner to operate independently of external data acquisition hardware, offering plug-and-play ethernet connectivity and immediate access to thermally corrected, calibrated engineering unit data.

Developed based on customer feedback and field performance data, the MPS4264 Gen2 incorporates the same advanced architecture found in the MPS4216 and MPS4232 models. The updated design enables fully synchronous scanning, achieved through one dedicated analog-to-digital converter (ADC) per sensor, eliminating inter-channel latency and offering competitive performance in time-critical applications. Sampling rates have also been increased to capture faster dynamic measurements with greater precision.

The onboard web server has been refined for ease of use, enabling quicker setup and real-time data access from any standard web browser. In addition, Scanivalve has developed a new published accuracy specification – ‘typical measurement error’. This metric is derived from 10 years of calibration data across thousands of MPS units and provides a grounded representation of real-world performance.

With this release, the entire MPS family – MPS4216, MPS4232 and MPS4264 Gen2 – now operates on a unified hardware and firmware platform. This ensures consistent performance, simplified updates and streamlined support across all models, with the MPS4216 and MPS4232 also receiving accuracy and usability enhancements as part of the update.

In related news, Evolution Measurement updates ProCap aerodynamics system

More on the latest sensor technology in the November 2025 edition of ATTI

As electric vehicles remove the masking effect of an internal combustion engine, the cabin’s true acoustic signature becomes far more noticeable. Drivers suddenly hear it all: tire and aerodynamic noise, the hum of auxiliary systems, the rumble of drivetrain components, and the high-frequency tonal ‘whines’ produced by electric motors and inverters. These tones occur at very low sound pressure levels, yet they significantly affect overall sound quality. The shift raises a fundamental challenge for NVH testing engineers: how do you measure extremely low sound pressure level (SPL) at high frequencies without the microphone itself introducing distortion?

Selecting the right microphone typically involves weighing competing performance factors. Measuring low SPL demands high sensitivity and a very low noise floor – conditions traditionally met only by a ½in microphone. However, a microphone of that size introduces substantial diffraction errors at high frequencies. Diffraction, or interference in a sound field, becomes problematic when the microphone diameter approaches the wavelength of sound – a condition that occurs increasingly as frequency rises. These errors depend on the characteristics of the sound field, the size and orientation of the microphone and the frequency of interest. Even small deviations from the ideal alignment can cause significant measurement inaccuracies.

To minimize diffraction-related errors, microphones are engineered for specific sound fields: pressure, random incidence (diffuse) and free field. Each design aims to maintain constant sensitivity or a flat frequency response when the microphone is aligned correctly. For free-field microphones, this means pointing directly at the sound source in an environment free of reflections. When the microphone is oriented away from 0° incidence, sensitivity at high frequencies is reduced. This is where the physics becomes limiting for traditional NVH tools.

A ½in free-field microphone such as PCB model 378B02 is a highly dependable acoustic measurement sensor, with low-power ICP (integrated electronics piezo-electric [IEPE]) operation, excellent stability, a low noise floor and 50mV/Pa sensitivity. Its larger size, however, makes it susceptible to diffraction and orientation errors at higher frequencies. In an effort to reduce diffraction, engineers may opt for a smaller ¼in free-field microphone such as the PCB model 378C01. However, its higher 42dB(A) SPL typical noise floor and lower 2mV/Pa sensitivity limit its ability to capture the low-level tonal noise that EVs reveal so clearly.

This long-standing compromise between accuracy and noise performance is exactly what PCB’s new model 378A08 was developed to overcome. With a noise floor of just 22dB(A), it is the quietest ¼in measurement microphone in the world. Its compact size dramatically reduces diffraction errors, while its high sensitivity enables accurate capture of low SPL at high frequencies. The microphone also retains all the advantages of ICP technology, including low-power two-wire connectivity and the environmental stability that PCB microphones are known for.

The benefits of this design become clear when examining how ¼in and ½in microphones behave in different acoustic conditions. In Figure 1, the deviation of microphone sensitivity from its calibrated value is shown for several orientations. This deviation represents measurement error. Both microphone sizes exhibit some attenuation in sensitivity as frequency increases, depending on orientation, but the reduction is far more pronounced for the ½in microphone. The ¼in microphone maintains much more consistent sensitivity, making it a more reliable choice when the sound field or microphone direction cannot be perfectly controlled.

The noise floor behavior shows a similar pattern. A microphone’s free-field noise specification is published for 0° incidence, but in real applications, microphones often operate at different angles or in mixed field types. Larger microphones experience an increase in noise floor under these conditions due to diffraction. As shown in Figure 2, the small diameter of the 378A08 allows it to maintain a low noise floor even when oriented away from 0°. Remarkably, at 120° incidence and frequencies above 13.5kHz, its noise is actually lower than that of a ½in microphone.

Together, these results position the 378A08 as a practical solution for applications that require both reliable high-frequency accuracy and exceptionally low noise, particularly in the low-level tonal environments now common in EV NVH testing.

At the core of the system is a modular rotor unit that mounts directly onto the driveshaft. Depending on the application requirements, the sensor can transmit measurement data either through a 2.4GHz radio telemetry link or via an inductive near-field interface. The radio variant provides a proprietary, interference-resistant wireless signal through FCC correlations with sampling rates of up to 10kHz and a transmission range of up to 10m, making it ideal for on-road applications. For long-duration or high-dynamics applications – especially on engine, e-motor or driveline test benches – the inductive version supplies continuous power across a typical rotor-stator distance of around 10mm. This enables uninterrupted data acquisition even under demanding thermal and mechanical boundary conditions.

The ME-DST’s electronics work across extended temperature ranges and are housed in sealed enclosures rated up to IP65/67. The rotor module can be customized for different shaft diameters and shapes, making it possible to fit the system into tight or unconventional spaces. Inside, the measurement electronics process strain-gauge signals with high stability, ensuring accurate torque readings even during rapid load changes or dynamic torsional oscillations.

To facilitate seamless integration into existing measurement infrastructures, the ME-DST receiver has analog and digital output interfaces. The analog output can be configured to deliver common current ranges such as 4-20mA, enabling direct connection to conventional data acquisition systems without the need for external converters. For more advanced applications, the system provides a fully configurable CAN interface in which message identifier, bitrate and transmission frequency can be adjusted to match the vehicle or test bench architecture. The torque value is transmitted as a high-resolution, processed signal, ensuring compatibility with modern control systems and data loggers.

A central component of the ME-DST’s user experience is its integrated web-based configuration interface. Once powered, the receiver creates a dedicated wi-fi network through which engineers can connect using any browser-enabled device. The interface provides a clear dashboard with live torque data, system health indicators, temperature information and, when applicable, inductive power levels. In the Live Data view, measurement curves can be monitored in real time with an auto-scrolling chart that supports pausing and data clearing for quick analysis during setup. The configuration menu facilitates rapid adjustment of output parameters, sampling rates and communication settings, while a simple tare function enables immediate zeroing of the sensor after installation. This modern UX-oriented design significantly reduces setup times and eliminates the need for proprietary software tools.

According to Melectric Systems, the ME-DST is suitable for a range of applications. In electric powertrain development, it enables precise mapping of motor torque, load cycles and efficiency under real operating conditions. In hybrid powertrains, detailed measurements during transient load changes, clutch engagements and regenerative braking events are beneficial. The system is also well suited to transmission testing, where its stable high-speed sampling can be used to analyze torsional vibrations, gear efficiency and lubrication performance. In motorsport and high-performance vehicles, the ME-DST’s low mass, compact size and robust telemetry make it ideal for real-time driveline analysis. For durability and endurance testing, the inductive version provides continuous operation without maintenance interruptions.

In the latest edition of ATTI, discover how suppliers are tackling today’s sensor R&D challenges

In related news, Metis has unveiled its next-gen eight-channel thermocouple module for harsh environments

Robustly estimating the structure of the world and ego-motion of the vehicle has been a multi-decade pursuit in computer vision. Classical feature-based and photogrammetry methods have long defined the state of the art in multi-view geometry. However, recent advances have shown that large-scale, transformer-based learning can push this frontier even further. Still, these methods fall short in the autonomous vehicle setting, where multiple cameras continuously capture the world through synchronized, structured rigs.

Rig3R is the first learning-based method to explicitly use multi-camera rig constraints for accurate and robust 3D reconstruction. It achieves state-of-the-art performance in complex, real-world driving scenarios.

The solution extends geometric foundation models with the advantage of rig awareness, leveraging rig information when available and inferring rig structure and calibration when it is not. This flexibility is essential for handling the diverse and evolving sensor setups found in embodied AI systems.

How Rig3R works

Rig3R is a machine learning model that takes images from multiple cameras and builds an accurate 3D understanding of the world. A large Vision Transformer (ViT-Large) processes each input image, breaking it into small patch tokens using 2D sine-cosine positional embeddings.  Each patch is enriched with a compact metadata tuple: Camera ID, Timestamp and a 6D Raymap, which encodes the camera’s pose relative to the rig. These fields provide spatial and temporal context, helping the model reason across multiple views and time.

A second ViT-Large decoder attends to all images jointly, across views and time, merging visual features, metadata and geometric cues into a shared latent space. During this fusion stage, Rig3R introduces a rig encoder to inject known rig constraints, enabling geometry-aware multi-view reasoning. The model is trained to leverage rig metadata when available but remains robust even when such information is missing. This fused representation forms the core of Rig3R’s multi-view 3D understanding.

Benchmarks and setup

Rig3R is evaluated on two multi-camera driving benchmarks: the Waymo Open validation set and WayveScenes101. Waymo provides lidar-based ground-truth, while WayveScenes101 uses COLMAP reconstructions. In the version presented in the recent paper, Rig3R is trained only on the Waymo training split, making WayveScenes101 an out-of-distribution test that evaluates generalization to unseen camera rigs.

Both datasets use five-camera rigs capturing approximately 200 frames per scene at 10fps. For evaluation, two 24-frame clips per scene are extracted, spaced approximately two seconds apart. Rig3R is benchmarked against feed-forward baselines, classical structure-from-motion and rig-aware methods.

The first objective, pose estimation, is assessed using relative rotation accuracy (RRA) and relative translation accuracy (RTA) at 5° and 15° thresholds, as well as mean average accuracy (mAA) up to 30°. The second objective, 3D pointmap reconstruction, is evaluated using accuracy, completeness and chamfer distance.

Results

The figure above compares pose estimation and pointmap reconstruction across four methods — Fast3R, DUSt3R-GA, Rig3R (unstructured) and Rig3R (with rig constraints) — on the same driving scene. The progression shows how Rig3R’s rig awareness improves both geometric coherence and reconstruction quality: while baseline methods often produce noisy or spatially inconsistent pointmaps and poses, Rig3R yields sharper, more consistent structures with rig poses that align sensibly across all views.

Why metadata matters

As more metadata is introduced, both pose estimation and pointmap reconstruction improve: predicted camera trajectories progressively align with the physical rig and reconstructed pointmaps become sharper and more consistent. This analysis illustrates how such structured priors, often readily available in embodied and robotic systems, can be effectively integrated into a learned geometric model to enhance both accuracy and generalization.

Rig3R achieves strong pose estimation and dense 3D reconstruction across diverse rig setups and conditions, including changes in baseline, field of view, lighting, speed and weather. It produces stable, low-drift trajectories and metrically consistent pointmaps, even in challenging scenes and in-the-wild driving videos — maintaining performance despite partial or missing metadata.

Many open problems remain in the quest to build a spatially intelligent foundation model. In addition to scaling up training, future improvements to Rig3R could include streaming representations, handling scene motion, multimodal inputs, multiple embodiments and multi-task outputs.

This is an edited version of an article that first appeared on Wayve’s blog on October 15, 2025.

In related news, Aeva introduces open-access FMCW 4D lidar and camera dataset for autonomous vehicle research

The Future of Automotive Testing Conference

This day of high-level content will explore various topics and provide an engaging opportunity to learn from and connect with industry peers. Situated among the exhibits, it will have a relaxing atmosphere, and visitors will be able to dip in and out at their leisure. A highlight will be ATTI’s on-stage interview with Jon Quigley, who worked for Volvo for many years and now runs his own company, Value Transformation. With his broad experience and portfolio of contacts, he’s a very good person to know. He loves sharing his thoughts, too, hence his new gig as ATTI’s regular columnist.

Read Quigley’s latest column on p36 of the September issue.

Also during the conference, Venkat Adusumalli, software engineering manager at Stellantis, will reveal how Stellantis is using AI to accelerate E/E system testing and validation workflows. Click here for a pre-show interview with him.

Innovation Showcase

The free-to-attend Innovation Showcase also provides an interactive platform for learning about new technologies. Speakers will welcome questions and discussion after their presentations, making it a great way to explore ideas together. Like the conference, it will take place within the main hall, so it will be very easy to listen in while wandering through the exhibits, or to come and go throughout the day.

Technology debuts

The industry is moving quickly and it can be a challenge to keep up. Every year there is an abundance of new technologies on display. Some are incremental improvements to existing products, others are entirely new. It’s always surprising how companies find ways around bottlenecks. Often, these developments emerge from ongoing conversations between supplier and customer, ultimately benefiting multiple companies.

Read ATTI‘s comprehensive expo preview in the September issue.

Face-to-face discussions  

It sounds obvious, but the most fruitful conversations are usually the ones that happen in person – it’s how journalists find the best stories. Year after year, exhibitors and visitors report having forged unexpected collaborations at the show, in the form of technology tie-ups or service partnerships.

Detroit’s revival 

A lot of money has been plowed into Detroit in recent years, as evidenced by the stunning renovation of Michigan Central Station, which is well worth a visit. Much of the investment has been in the tech sector. Eleven years ago – when the station was still derelict – ATTI took a tour of some of the labs at GM and what was then the FCA headquarters (see The right wave length, November 2014, Seasonal highlights, September 2015, and Powerbase, March 2015). The facilities were impressive then and are even more so now, thanks to several years of investment in innovation leading to Detroit’s resurgence as an automotive powerhouse.

Visit the Automotive Testing Expo North America website to secure your FREE pass, which will give you access to the expo, the Future of Automotive Testing Conference and the Innovation Showcase