FAQs on Darsi Pro Compute Box

Darsi™ Pro is an Edge AI compute box based on the NVIDIA® Jetson Orin™ NX, suitable for mobility and AI vision applications. This production-ready compute platform supports multiple sensors, including cameras, LiDAR, and radar, and provides reliable hardware for edge computing.

The Darsi™ Pro compute box supports e-con Systems' broad range of GMSL cameras, Ethernet cameras and USB cameras.

The Darsi™ Pro compute box supports an input voltage of 24V, accommodating various vehicle battery configurations. Based on the hardware requirements, it can be designed from range of 8V to 36V DC. The system incorporates comprehensive protection circuitry, including reverse polarity protection, input surge protection, inrush current limiting, and battery-percentage-based power management.

The Darsi™ Pro compute box typically consumes 25W.

The power management MCU serves as the main controller for shutdown, soft reboots, sleep and wake-up functions, and power operating modes. Furthermore, in the event of a failure, the power management MCU acts as a watchdog to guarantee appropriate recovery.

The system has two USB 3.0 ports, allowing up to 5 Gbps speeds for data transfer.

M.2 connectors are available for modular expansion. The NVMe SSD utilizes four dedicated PCIe lanes, while other slots support Wi-Fi modules and LTE/5G connectivity modules based on specific application needs.

The vision box includes three 1Gbps Ethernet ports—two supporting Power over Ethernet (PoE) and one standard Ethernet port without PoE.

The system includes three CAN interfaces designed for integration with vehicle control systems.

The compute box is designed to support sensor fusion across LiDAR, RADAR, and camera systems. The use of the GNSS module is being assessed to provide a PPS (Pulse Per Second) master signal generator for sensor synchronization. The implementation strategy includes attaching trigger timestamps to the framebuffer to ensure precise time synchronization across all connected sensors.

Read Blog: What Sensor Fusion Architecture Offers for NVIDIA Orin NX-Based Autonomous Vision Systems - e-con Systems

Both eSIM and physical SIM slots are provided to connect external antennas for optimal signal reception.

Yes, there is an Inertial Measurement Unit (IMU) to determine the spatial orientation and motion characteristics.

The compute box runs a Linux image built on NVIDIA's L4T (Linux for Tegra) BSP for NVIDIA® Orin™ NX module series. This approach provides long-term support and production-grade reliability.

CloVis Central™ is a device management platform based on the cloud and acts as a unified platform for feature updates, system upgrades, and fleet maintenance.

The architecture supports secure Over-The-Air updates for both the base system and containerized applications. Integration with third-party update services such as Azure IoT is under consideration.

The compute box uses the NVIDIA® Secure Boot built into the Jetson platform hardware. All bootloader and kernel images are encrypted for prevention against unauthorized or compromised firmware execution. A hardware-based UUID is optionally used to enable authentication of the device.

Read Blog: What is Secure Boot and How Does it Safeguard Edge AI Vision Deployments? - e-con Systems

Remote access is secured through password-based or key-based authentication. The system includes provisions for integrating TPM version 2.0 or equivalent secure components in future versions.

Device lockdown is implemented through multiple security mechanisms:

• Root shell access is disabled in production configurations.
• Software reflashing can be performed only by authenticated users.
• A fail-safe mode ensures the system can recover to a secure operational state during critical failures or security incidents.

• MCU-based hardware watchdog to detect and recover from system hangs or boot faults
• Software health monitoring with recovery mechanisms for cameras
• Controlled power sequencing
• Critical event logging for diagnostics
• Optional GPIOs or error pins signaling for external health monitoring

Software validation encompasses extended runtime stability testing (soak test/burn-in), boot/reboot cycle validation, watchdog and fault recovery verification, peripheral stress testing, and CPU/GPU stress testing to ensure system robustness under all operational conditions.

The vision box has dimensions of 326 x 179 x 110 mm in a rugged box PC form factor. The enclosure features an aluminum body with gasketed ends for IP protection.

The system is designed to achieve IP67 certification, providing protection against dust ingress and temporary water immersion.

The vision box complies with FCC Part 15 Class A and CISPR25 standards. Environmental compliance includes RoHS 3 and REACH requirements.

The product targets compliance with ISO16750 road vehicle environmental conditions and testing standards, along with IP67 ingress protection, ensuring suitability for automotive and industrial mobile applications.

The validation includes thermal cycling tests, vibration and mechanical shock testing as per road vehicle standards, and comprehensive EMC (Electromagnetic Compatibility) and ESD (Electrostatic Discharge) testing.

The product includes a 5-year software update lifecycle, covering security patches, feature updates, and platform maintenance.