Sonair’s Dolphin-Inspired Ultrasound Tech Paves Way for Lidar-Free 3D Vision in Autonomous Systems

Sonair, a Norwegian startup, is taking a novel approach to 3D computer vision by drawing inspiration from dolphins. Instead of relying on traditional lidar technology, which can be expensive and power-hungry, Sonair is leveraging ultrasound to create an efficient and cost-effective solution for autonomous systems.

What Changed Technically

Sonair's innovation lies in its use of ultrasound for 3D mapping and object detection. Ultrasound has traditionally been used in medical imaging and parking sensors due to its non-invasive nature and low power consumption. However, Sonair is pushing the boundaries by applying it to more complex tasks typically handled by lidar.

• Ultrasound vs. Lidar:
  • Range and Accuracy: While lidar offers high precision over long distances, ultrasound can achieve comparable accuracy within shorter ranges (up to a few meters). This makes it ideal for indoor and close-range applications.
  • Energy Efficiency: Ultrasound sensors consume significantly less power than lidar systems, making them more suitable for battery-powered devices like drones and robots.
  • Cost-Effectiveness: The components required for ultrasound are generally cheaper and more widely available than those for lidar.

How It Works

Sonair's system works by emitting high-frequency sound waves and measuring the time it takes for these waves to bounce back after hitting an object. This data is then processed to create a 3D map of the environment. The key technical challenges and solutions include:

• Signal Processing:

  • Noise Reduction: Advanced signal processing algorithms are used to filter out background noise and enhance the quality of the received signals.
  • Multi-Frequency Emission: Using multiple frequencies helps in achieving better resolution and reducing interference from other sources.

• Data Fusion:

  • Combining Data Streams: Sonair's system can integrate data from multiple ultrasound sensors, as well as other sensor types like cameras and inertial measurement units (IMUs), to create a more robust and accurate 3D model.
  • Real-Time Processing: The system is designed for real-time operation, allowing it to quickly adapt to dynamic environments.

Implementation Details

• Hardware:

  • Transducers: Sonair uses high-sensitivity transducers that can both emit and receive ultrasound waves. These are arranged in a specific pattern to cover the desired field of view.
  • Microcontrollers: Low-power microcontrollers handle the signal processing and data fusion tasks, ensuring efficient operation.

• Software:

  • Firmware: Custom firmware is developed to manage the transducers and process the raw data into usable information.
  • Algorithms: Machine learning algorithms are employed to improve object recognition and environment mapping.

Benchmarks and Performance

Early tests have shown promising results:

• Accuracy: The system can achieve sub-centimeter accuracy within a range of up to 5 meters.
• Power Consumption: It consumes less than 10% of the power required by a typical lidar system.
• Latency: Real-time processing ensures that the system can update its environment map in milliseconds.

Use Cases

Sonair's technology has a wide range of potential applications:

• Indoor Navigation: Drones and robots can navigate complex indoor environments with high precision.
• Quality Inspection: The system can be used for non-destructive testing and inspection in manufacturing settings.
• Healthcare: It could assist in medical imaging and robotic surgery, where precise 3D visualization is crucial.

Future Outlook

Knut Sandven, Sonair's founder and CEO, believes that ultrasound-based 3D vision will play a significant role in the future of autonomous systems. "By combining the efficiency and cost-effectiveness of ultrasound with advanced signal processing, we can create solutions that are not only more accessible but also more sustainable," he said.

Conclusion

Sonair's approach to 3D computer vision using ultrasound is a promising alternative to lidar. With its energy efficiency, cost-effectiveness, and potential for real-time operation, it could open up new possibilities in various industries where precise 3D mapping is essential.