DaischSensor’s IMU Motion Sensing Solution for Autonomous Driving

Autonomous Driving Sensors are one of the most important components in the intelligent driving. On September 12, SENSOR CHINA 2024 was held in Shanghai. It is one of the world’s top three sensor exhibitions. DAISCH co-founder and CEO Jiang Bian presented a keynote entitled Motion Sensing Solutions for Automotive Intelligent Driving and Dynamic Testing Applications on this grand stage of cutting edge technologies and future smart industries during this forum on autonomous driving sensors technology and application. Let’s revisit the highlights of Mr. Bian’s insightful presentation by focusing on three key questions:

1. In the evolution of self driving, why has the IMU (inertial measurement unit (IMU)) become an indispensable autonomous driving sensors?
2. As we go from lower to higher level self driving, what are the challenges for IMU?
3. How does DAISCH IM8 provide the “perfect solution” for advanced self driving?

Part 1: IMUs Become Key autonomous driving Sensors in Intelligent Driving

Motion sensing components of traditional vehicle sensor include, of course, wheel speed sensors, yaw rate sensors, collision sensors, suspension acceleration sensors, suspension displacement sensors, satellite positioning, and so on.

As self driving grows, cameras, LiDAR, and high precision, high-precision 6-axis IMUs are becoming more important in sensing the environment and the vehicle.

Of these, the high precision IMU is the only sensor that does not emit or receive external signals, and operates autonomously.

In recent years, as lower-level classic self driving advances toward higher-level end-to-end self driving, the vehicle motion sensing solution has evolved from a GNSS module and PBOX to an independent IMU module integrated within a domain controller. This demonstrates that the IMU will take on an even more significant role in future intelligent driving scenarios.

In classic intelligent driving applications, the use of IMUs has become well-established, commonly used in the following modules and functions.

• Vehicle Positioning and Navigation: With GNSS location data or other sensors such as LiDAR, cameras, and odometers, the IMU can supplement positioning accuracy and reliability by adding measurement of acceleration and angular velocity measurement.
• Motion State Perception: Realties of sensing, capability is that the vehicle’s acceleration and angular velocity can be measured in real time by the IMU and so the system can determine the vehicle’s motion status and adjust power output and braking strategies to guarantee safe driving.
• Environment Perception Assistance: The vehicle’s acceleration and angular velocity changes as presented by the IMU are supplemented with the data from other sensors to help the system quickly identify and track dynamic objects around the vehicle.  The IMU also assists in map matching, improving positional accuracy.
• Control and Decision Support: The IMU supplies the system with the vehicle’s current position, velocity, and acceleration to give the system some ideas for what the best driving path would be when the system has to choose the most driveable path.

Part 2: IMU in End-to-End Intelligent Driving Solutions

“End-to-end” in intelligent driving is much like the large language model technology behind ChatGPT. It is a full workflow that process raw sensors (cameras, radar, LiDAR) to final control commands (acceleration, braking, steering). On the other hand, there is a large variance on how the IMU can be applied in end to end and non end to end systems.

• Different Data Fusion Approaches: In end-to-end systems, IMU data is fused directly with other sensors (e.g., cameras, LiDAR) in a deep neural network to optimize driving decisions. In non-end-to-end systems, IMU data undergoes traditional signal processing and algorithmic modules before being integrated with other sensor data.
• Different Impact on Decision-Making: In end-to-end systems, IMU data directly influences driving decisions as the system uses deep learning models to connect sensor inputs with control outputs, making the IMU’s vehicle motion data critical to the model’s real-time outputs. In non-end-to-end systems, the IMU plays a supporting role, providing information to various system modules but having a more complex and indirect effect on the final decision.
• Explainability Differences: The role of the IMU in end-to-end systems is harder to explain due to the black-box nature of deep learning models. Although system output changes can be observed after IMU data is input, it’s difficult to precisely understand how the model uses IMU data to make decisions. In non-end-to-end systems, the IMU’s function is easier to analyze, as its data is processed in distinct modules and can be interpreted using traditional algorithms and engineering methods.

Part 3: Advanced IMU Requirements for End-to-End Intelligent Driving Solutions

It’s worth noting that in the increasingly popular end-to-end intelligent driving solutions, IMU-integrated domain control solutions have become the preferred choice for 80% of self driving clients, which presents higher challenges for IMUs.

• High Precision and Stability: A high-precision IMU sensor provides accurate vehicle attitude information, such as pitch, roll, and yaw angles, helping the system better understand the vehicle’s spatial position and orientation. Stability is critical for system reliability—IMUs must consistently deliver reliable data over long durations, unaffected by environmental factors like temperature changes and vibrations.
• Fast Response and High Update Frequency: End-to-end systems need to respond quickly to environmental changes, so IMUs must have fast response capabilities and high update frequencies. When the vehicle undergoes sudden accelerations, decelerations, or turns, the IMU must promptly detect these changes and provide data to the system for real-time decision-making.
• Compatibility with Deep Learning Models: End-to-end intelligent driving systems typically use deep learning models to process sensor data. IMUs need to provide data in a format compatible with these models, possibly requiring specific interfaces or software libraries to facilitate integration.
• High Reliability and Fault Detection: In intelligent driving systems, Safety is paramount, which means that IMUs need to operate reliable even in harsh conditions (such as extreme temperatures, high humidity, severe vibrations). IMU also shall have fault detection and diagnostic abilities allowing system to detect faults and take corrective course of action such as switching to alternate sensors, or reducing vehicle speed.

Part 4: DAISCH IM8 for Intelligent Driving Applications from Low-Level to High-Level

• Optimal SWaP (Size, Weight, and Performance): Limited space inside domain controllers demands miniaturization and lightweight sensors. DAISCH IM8 uses UltraCompact technology to achieve the best SWaP while maintaining a miniature size.
• Microsecond-Level Response: The IM8 inertial module provides microsecond-level response and updates at 1000Hz, allowing it to rapidly detect vehicle dynamics and provide timely data to the system, assisting in making appropriate decisions.
• ASIL-B Functional Safety Certification: During the development and design of the IM8, functional safety requirements were fully considered. The product follows the ISO 26262 standard, earning ASIL-B functional safety certification.
• AutoCalix Technology for Mass Production Consistency: IM8 adopts DAISCH’s AutoCalix technology to calibrate and correct errors for large-scale IMU products, ensuring that every IM8 delivered to customers meets the requirements of self driving systems. It also features self-learning and fault diagnosis capabilities.

Conclusion

From low-level to high-level intelligent driving, DAISCH IM8 excels with high precision, lightweight design, and high reliability, comprehensively meeting the needs of self driving solutions and driving the acceleration of a smart future. Economically, DAISCH’s innovative IM8 domain control integration solution creates annual value in the tens of millions for clients, reducing system-level costs by over 50% compared to previous solutions. Since its mass production began in 2023, DAISCH IM8 has become the top choice for nearly ten automotive OEMs and Tier 1 suppliers across almost 20 vehicle models.

Beyond intelligent driving, DAISCH IMUs also play indispensable roles in robotics, low-altitude economy, smart agriculture, and mining machinery. With the advent of the intelligent era, DAISCH IMUs will see even broader market prospects.