# Prism-A4L 413GSV0 Product Specification

Prism-A4L 413GSV0 supports **assembly and acquisition with 1–4 cameras**: one, two, three, or four cameras can be used as needed; all four are not required. Assemble the cameras from the standard four-camera kit to suit the application. Wiring, preview, recording, and acceptance checks apply to the cameras actually connected and enabled. Four-camera diagrams, videos, and checks illustrate the full four-camera configuration.

[Body ports & LED indicators](?doc=panels&lang=en)


## Specification summary

| Category | Prism-A4L 413GSV0 specification |
|---|---|
| Cameras | 1–4 × SC130GS global-shutter cameras |
| Image output | Current: 1–4 × 1280 × 1024 MJPEG; future support: 4 × 2 MP global-shutter cameras |
| Camera synchronization | Shared trigger reference for all four cameras |
| Frame-rate configuration | Any integer from 1 to 30 fps; 30 fps default |
| Image controls | MJPEG Quality, exposure range, gain range, target brightness |
| Onboard IMU | One standard; up to two on designated product variants |
| Onboard IMU data | Acceleration, angular velocity, temperature, timestamp, synchronization status |
| Optional LiDAR | Livox Mid-360 / Mid-360S / Hesai XT32 |
| LiDAR data | Point cloud; internal LiDAR IMU where provided by the model (not available on XT32) |
| Data time | Unified device time base; hardware-level external PPS input synchronization supported, with valid NMEA time messages required for external UTC synchronization |
| Host connection | USB 3.x |
| Power | Standby 2.5 W; data transfer 4.5 W |
| Preview | 4 Cameras, onboard IMU, point cloud, LiDAR IMU |
| Recording modes | Full, IMU-only |
| Data format | prism-dataset-v6 |
| Data checks | Integrity, index range, required streams, timestamp anomalies |
| Data playback | Local dataset playback by time relationship |
| Application integration | Host SDK |
| ROS support | ROS1/ROS2 adaptation; ROS1 Bag and ROS2 rosbag2 SQLite3 export |

### Notes

Except for the explicitly marked future support for four 2 MP global-shutter cameras, all capabilities listed in this specification are currently supported by Prism-A4L 413GSV0, including hardware-level external PPS time synchronization. Optional LiDAR, dual onboard IMUs, and related functions apply only to configurations containing the corresponding hardware.

## 1. Document information

### 1.1 Purpose

This specification defines the currently supported product functions, data types, workflows, technical specifications, and delivery acceptance requirements for Prism-A4L 413GSV0. It also identifies capabilities planned for future support and is intended for product selection, solution review, field deployment, and delivery acceptance.

This document does not describe internal software architecture, board-level implementation, or private interfaces. Items marked “future support” are outside the scope of current delivery acceptance.

## 2. Product overview

### 2.1 Product positioning

Prism-A4L 413GSV0 is a multi-sensor synchronized acquisition device for robotics, mapping, spatial perception, calibration, data collection, and algorithm development. It supports assembly with 1–4 global-shutter cameras and integrates an onboard six-axis IMU, with optional Livox Mid-360, Mid-360S, or Hesai XT32 LiDAR. Image, inertial, and point-cloud data are acquired on a unified device time base, with Host SDK, ROS1, and ROS2 integration support.

### 2.2 Core capabilities

- Acquisition from 1–4 1280 × 1024 global-shutter cameras, using a shared trigger reference for multi-camera configurations;
- Future support for four 2 MP global-shutter cameras;
- Any integer frame rate from 1 to 30 fps;
- Camera MJPEG quality, exposure, gain, and target-brightness settings;
- One onboard six-axis IMU as standard, with an optional dual-onboard-IMU configuration for a total of two;
- Optional Livox Mid-360, Mid-360S, or Hesai XT32 point-cloud acquisition; internal LiDAR IMU data is model-dependent and is not provided by XT32;
- A unified device time base for cameras, onboard IMUs, and LiDAR data;
- Hardware-level external PPS time synchronization, with valid NMEA time messages establishing an external UTC time base;
- Real-time data transfer over USB 3.x;
- Real-time camera, IMU, and point-cloud preview;
- Full and IMU-only recording modes;
- Dataset integrity and timestamp checks;
- Local dataset playback;
- Host SDK application integration;
- ROS1/ROS2 adaptation, plus ROS1 Bag and ROS2 rosbag2 export.

### 2.3 Standard and optional components

| Component | Quantity | Description |
|---|---:|---|
| Prism-A4L 413GSV0 main unit | 1 | Multi-sensor acquisition, synchronization, and USB data output |
| SC130GS global-shutter cameras | 4 | Four-camera shared-trigger synchronization |
| Onboard six-axis IMU | 1 | Standard configuration, ICM-45686 by default |
| Dual onboard IMU configuration | 2 | Optional and available only for the corresponding product variant |
| Livox Mid-360 / Mid-360S / Hesai XT32 | 0 or 1 | Optional by purchase configuration |

Device status and dataset content are generated from the actual hardware configuration. No false stream is created for an uninstalled second onboard IMU or an unconnected LiDAR.

## 3. Typical workflow

1. Connect Prism-A4L 413GSV0 to the host USB 3.x port.
2. Open Prism Viewer and connect to the device.
3. Check the operating status of the four cameras, onboard IMU, and optional LiDAR.
4. Configure camera frame rate, MJPEG quality, exposure, and gain.
5. If LiDAR is installed, configure its model and network parameters.
6. Synchronize device time and check the result.
7. Start real-time preview or data recording.
8. After recording stops, check dataset integrity and timestamps.
9. Play back the dataset or export ROS data as needed.

## 4. Camera functions and specifications

### 4.1 Basic specifications

| Item | Specification |
|---|---|
| Camera count | 1–4, assembled and acquired as needed |
| Image sensor | SC130GS |
| Shutter type | Global shutter |
| Output resolution | 1280 × 1024 |
| Output format | MJPEG |
| Synchronization | All four cameras use the same trigger reference |
| Frame-rate range | Any integer from 1 to 30 fps |
| Default frame rate | 30 fps |
| Default MJPEG Quality | 88 |
| Default gain | 1× |
| Default target brightness | 35 |
| Minimum exposure time | 50 µs |

### 4.2 Frame-rate and image-quality settings

Users can set any integer frame rate from 1 to 30 fps and set MJPEG Quality through the Host SDK or Prism Viewer. Settings apply to all four cameras, which continue synchronized acquisition using their shared trigger reference.

Increasing MJPEG Quality increases per-frame data size and device encoding, USB transport, and host decoding load. If MJPEG quality is set too high, high-FPS operation may experience a lower sustained frame rate or increased response latency. When high frame rate and real-time response are the priority, use the default quality of 88 or reduce MJPEG Quality as appropriate. Higher quality settings should be validated for sustained frame rate with the actual host, LiDAR configuration, and acquisition load.

30 fps is a configurable operating point. Actual sustained frame rate also depends on the host USB controller, host load, LiDAR data volume, and device temperature. Delivery acceptance should use sustained acquisition testing with the final product configuration.

### 4.3 Exposure and gain control

Prism-A4L 413GSV0 supports minimum and maximum exposure, minimum and maximum gain, and target-brightness settings. The minimum exposure time is 50 µs. Current automatic brightness adjustment primarily uses an exposure-time closed loop while preserving the fixed gain set by the user.

Exposure time is limited by the current frame period. The valid upper limit is calculated as follows:

`maximum exposure time ≤ floor(1,000,000 / fps) - 5,000 µs`

Users can write exposure settings before acquisition starts and can update permitted exposure parameters while acquisition is running.

### 4.4 Image color and highlight protection

The four cameras use shared automatic white balance based on actual image statistics and include protection against purple or white flashing caused by large highlights. Image-quality acceptance covers indoor lighting, outdoor daylight, mixed lighting, strong highlights, transitions between bright and dark scenes, and low-light scenes.

### 4.5 Image metadata and start/stop behavior

- Images from Camera 1–4 contain valid exposure metadata;
- Camera timestamps reference the shared trigger edge;
- Unstable startup frames are ignored, and the first externally visible frame belongs to the current acquisition session.

## 5. Onboard IMU

### 5.1 Configuration and data types

| Item | Specification |
|---|---|
| Standard quantity | One six-axis IMU |
| Optional configuration | Two onboard six-axis IMUs in total |
| Default device | ICM-45686 |
| Compatible devices | ICM-42688 or ICM-45686, depending on product configuration |
| Data types | Three-axis acceleration, three-axis angular velocity, temperature, timestamp, and synchronization status |
| ICM-45686 configuration | 800 Hz, ±8 g, ±1000 dps, approximately 200 Hz filtering |

### 5.2 Display units

Prism Viewer supports the following display units:

- Acceleration: mg, g, m/s²;
- Angular velocity: mdps, °/s, rad/s;
- Temperature: m°C, °C.

Changing display units does not alter recorded data. Dataset recording and ROS export use SI units: m/s² for acceleration, rad/s for angular velocity, and °C for temperature.

### 5.3 Multiple-IMU identification

The second onboard IMU appears only in the corresponding product variant. A single-IMU product displays and records only one onboard IMU stream; a dual-IMU variant displays and records the two onboard IMU streams separately.

## 6. Optional LiDAR

### 6.1 Supported models and connection

| Item | Specification |
|---|---|
| Supported models | Livox Mid-360, Mid-360S, Hesai XT32 (PandarXT-32) |
| Connection | Dedicated Prism-A4L 413GSV0 Ethernet interface |
| Configuration | Interface enable, device IPv4, netmask, LiDAR IPv4 |
| Common network example | Prism-A4L 413GSV0: 192.168.1.5/24; LiDAR: 192.168.1.3 |
| Data types | Point cloud; internal LiDAR IMU where provided by the model (not available on XT32) |

The network configuration page separately displays the parameter save result, LiDAR connectivity, and point-cloud reception state. A successful save means only that the parameters were written; it does not mean point clouds are being received.

Supports **Hesai XT32 (PandarXT-32) point-cloud acquisition and preview**. Connect through Ethernet and power the LiDAR separately. Select XT32 in the LiDAR settings and configure the network using the actual LiDAR address; stop acquisition before switching models. XT32 provides no internal LiDAR IMU stream; the Prism onboard IMU remains available. XT32 support does not imply support for XT32M2X or XT16.

### 6.2 Point-cloud preview

The Prism Viewer point-cloud window supports:

- Mouse rotation and zoom;
- Reflectivity coloring;
- Point-size adjustment;
- Top-down view;
- View reset.

### 6.3 Point-cloud and LiDAR IMU data

Point clouds are output in batches with a batch reference time. Intra-batch timing retains the source semantics from the LiDAR.

The LiDAR internal IMU and the Prism-A4L 413GSV0 onboard IMU are independent data sources and are identified separately in the interface, recordings, and ROS exports.

## 7. Time synchronization

### 7.1 Unified device time base

Cameras, onboard IMUs, LiDAR point clouds, and the LiDAR IMU use a unified device time base. Unified time means that the streams use the same time domain; it does not mean that different sensors produce samples at the same physical instant.

| Data stream | Time semantics |
|---|---|
| Camera | Shared trigger-edge time |
| Onboard IMU | IMU sample time; enters the unified device time domain when synchronization status is valid |
| LiDAR point cloud | Retains and normalizes the LiDAR PTP/TAI time source |
| LiDAR IMU | Retains the LiDAR time source and remains separate from the onboard IMU |

Acquisition uses device measurement time and does not substitute host USB arrival time for sensor time.

### 7.2 Host time synchronization

At startup or on the first successful device open, Prism Viewer automatically attempts to synchronize host and device time and reports success or failure. A synchronization failure does not prevent acquisition. Users can retry manually in the interface or initiate host-to-device time synchronization through the Host SDK.

The difference between LiDAR TAI and UTC is processed dynamically from valid time information rather than using a fixed offset in seconds.

### 7.3 External PPS time synchronization

Prism-A4L 413GSV0 currently supports hardware-level external PPS input synchronization through the TIMESYNC port, aligning its unified device time base to an external timing source for GNSS timing and multi-device synchronization.

- External UTC synchronization requires both **PPS pulses and valid NMEA RMC date/time messages**. PPS supplies the second edge and RMC supplies its absolute time label; PPS alone does not establish the UTC date and time.
- Select input mode and match the NMEA UART baud rate to the external timing device. See the [device interface guide](https://dibuli.com/docs.html?doc=panels&lang=en) for signal levels, wiring, and mode-switching requirements.
- PPS detected, valid PPS, valid RMC, and external UTC synchronized are distinct states. Confirm external UTC synchronization before use; detecting PPS alone is not sufficient.
- If the external timing signal is lost, the device continues on its internal time base. Continued operation does not mean it remains locked to external UTC; check synchronization status and timestamp validity flags when using the data.

Hardware-level external PPS input synchronization is included in current product support and delivery acceptance. Synchronization accuracy must be evaluated with the intended external timing source and installation.

## 8. Data connection and power consumption

### 8.1 USB data connection

Prism-A4L 413GSV0 uses USB 3.x to transport camera, onboard IMU, and optional LiDAR data. Full-load acquisition should use a host USB 3.x port.

One host application opens a given Prism-A4L 413GSV0 at a time. Network and recording-mode configurations are changed while data streams are stopped; camera exposure parameters can be updated during acquisition.

If the operating system denies USB access, Prism Viewer displays a device-access failure with permission diagnostics.

### 8.2 Power consumption

| Operating state | Power |
|---|---:|
| Standby | 2.5 W |
| Data transfer | 4.5 W |

## 9. Prism Viewer

### 9.1 Device and sensor operations

Prism Viewer provides:

- Device discovery, connection, disconnection, and status display;
- Four-camera layout, preview, start, and stop;
- Camera fps, MJPEG Quality, exposure, gain, and target-brightness settings;
- Onboard IMU display and unit switching;
- LiDAR model selection, network configuration, connectivity checks, and point-cloud reception status;
- Real-time point-cloud preview and view controls;
- Automatic device-time synchronization, result messages, and manual retry;
- Full and IMU-only recording;
- Local dataset browsing, integrity checks, timestamp checks, and playback;
- ROS1 Bag and ROS2 rosbag2 export.

The interface identifies onboard IMUs and the LiDAR IMU separately. Starting acquisition does not force the Viewer window size to match the original image dimensions.

### 9.2 Recording modes

| Recording mode | Without LiDAR | With LiDAR |
|---|---|---|
| Full | 4 Cameras + onboard IMU | 4 Cameras + onboard IMU + point cloud + LiDAR IMU |
| IMU-only | Onboard IMU | Onboard IMU + LiDAR IMU |

IMU-only mode creates only IMU data and does not create camera or point-cloud data files.

## 10. Data recording and checks

### 10.1 Dataset content

Prism-A4L 413GSV0 uses the `prism-dataset-v6` data format. Depending on the actual product configuration, a dataset contains:

- Four camera image streams and indexes;
- One onboard IMU, with a second onboard IMU in dual-IMU product variants;
- Optional LiDAR point clouds;
- Optional LiDAR internal IMU data;
- Dataset information, completion state, and sensor metadata.

### 10.2 Dataset integrity

After a recording ends normally, the dataset stores a completed state. Local dataset checks include:

- Data-format and file-structure checks;
- Completion-state checks;
- Index-reference and data-range checks;
- Checks for streams required by the current recording mode;
- Timestamp monotonicity checks for every stream;
- Checks for duplicate timestamps, rollback, and abnormal jumps.

A dataset that did not end normally is clearly marked as incomplete in the check results.

### 10.3 Dataset playback

Prism Viewer can open a local Prism-A4L 413GSV0 dataset and play back camera, onboard IMU, point-cloud, and LiDAR IMU data according to their time relationships. Playback shows only the streams actually contained in the dataset.

## 11. Software and ecosystem integration

### 11.1 Host SDK

Prism-A4L 413GSV0 supports integration into user applications through the Host SDK. The Host SDK provides these user-visible capabilities:

- Device enumeration, open, close, and basic-information reads;
- Reception of Camera, onboard IMU, LiDAR point-cloud, and LiDAR IMU data;
- Camera fps, MJPEG Quality, exposure, gain, and target-brightness configuration;
- LiDAR model and network-parameter configuration;
- Device time synchronization;
- Device status, data-reception status, and error-information reads.

The Host SDK delivery package includes public headers and runtime libraries for the corresponding platform. User applications should use mutually compatible components from the same delivery package.

### 11.2 ROS1/ROS2 adaptation and data export

| Prism-A4L 413GSV0 data | ROS adaptation and export |
|---|---|
| Camera | Image topics |
| Onboard IMU | Independent IMU topics using SI units |
| LiDAR point cloud | PointCloud2 |
| LiDAR IMU | Independent IMU topic using SI units |

Prism-A4L 413GSV0 supports ROS1 and ROS2 adaptation, allowing Camera, onboard IMU, LiDAR point-cloud, and LiDAR IMU data to connect to the corresponding ROS interfaces. Prism Viewer also exports ROS1 Bag and ROS2 rosbag2 SQLite3. ROS1 export targets the Noetic data format, while ROS2 export generates data readable by standard rosbag2 tools.

Exports preserve the independent identities of onboard IMUs and the LiDAR IMU as well as the device-time semantics of each data stream.

## 12. Current performance notes

### 12.1 Camera frame rate

Prism-A4L 413GSV0 supports integer frame rates from 1 to 30 fps. A Camera-only configuration can operate at the 30 fps setting with a suitable USB 3.x host and adequate cooling.

When Camera, LiDAR, and all IMUs are enabled together, actual camera frame rate is affected by the USB controller, host load, point-cloud traffic, and temperature. This combination supports synchronized acquisition, but continuously stable 30 fps in every environment is not a currently guaranteed metric.

### 12.2 Sustained acquisition

Delivery testing includes:

- Camera + onboard IMU continuous acquisition for 4 hours;
- Camera + LiDAR + all IMUs continuous high-load acquisition for 4 hours;
- Simultaneous logging of four-camera frame counts, IMU data, point-cloud data, USB timeouts, dropped frames or packets, and device temperature;
- Checking for unrecoverable stream stops or timestamp rollback during acquisition.

## 13. Usage requirements

- Use a USB 3.x connection for full-load acquisition;
- Place LiDAR network parameters in the same subnet as the actual LiDAR address;
- Check sensor status and time synchronization before formal recording;
- Process onboard IMUs and the LiDAR IMU separately when using the data;
- Run integrity and timestamp checks after data recording;
- Use the matching Prism Viewer from the delivery package and avoid mixing different releases;
- Provide sufficient sustained write bandwidth and available storage on the host.

## 14. Delivery acceptance criteria

The acceptance report records the product serial number, product configuration, Viewer release, host operating system, USB controller, LiDAR model, test duration, and environmental conditions.

| ID | Acceptance item | Pass criteria |
|---|---|---|
| AT-001 | Device connection | The device can be discovered, connected, and disconnected, and connection status is correct |
| AT-002 | Four cameras | All four 1280 × 1024 MJPEG streams can be previewed and recorded |
| AT-003 | Frame-rate setting | Integer settings from 1 to 30 fps are applied correctly |
| AT-004 | Four-camera synchronization | Four-camera frame groups can be associated, timestamps are monotonic, and no single stream drifts continuously |
| AT-005 | Exposure settings | Exposure, gain, and target-brightness settings are applied and can be updated while running |
| AT-006 | Exposure metadata | Exposure metadata is valid for Camera 1–4 |
| AT-007 | Image quality | Indoor, outdoor, strong-highlight, low-light, and lighting-transition scenes show no persistent abnormal flashing |
| AT-008 | Onboard IMU | Every installed IMU is identified, displayed, and recorded, with correct unit conversion |
| AT-009 | Single-IMU product | A nonexistent second onboard IMU is neither displayed nor generated |
| AT-010 | LiDAR configuration | Model and network parameters can be saved, and connectivity is distinguished from point-cloud reception |
| AT-011 | Point-cloud preview | Point clouds display continuously; rotation, zoom, coloring, point-size, and view controls work |
| AT-012 | LiDAR IMU | The LiDAR IMU is displayed, recorded, and exported separately from onboard IMUs |
| AT-013 | Time synchronization | Automatic synchronization messages are correct, and a failed attempt can be retried manually |
| AT-014 | Full recording | All camera, IMU, and optional LiDAR data is generated according to product configuration |
| AT-015 | IMU-only recording | Only onboard IMU and optional LiDAR IMU data is generated |
| AT-016 | Dataset checks | Completion state, file structure, indexes, data ranges, and required streams are checked correctly |
| AT-017 | Timestamp checks | Duplicate timestamps, rollback, and abnormal jumps can be detected |
| AT-018 | Data playback | Cameras, IMUs, and point clouds contained in a dataset play back according to their time relationships |
| AT-019 | Host SDK | The device can be enumerated and opened, cameras configured, and streams present in the product configuration received |
| AT-020 | ROS1/ROS2 export | Images, IMUs, and PointCloud2 are readable by the corresponding standard ROS tools |
| AT-021 | Camera + IMU endurance | 4 hours of continuous acquisition without an unrecoverable stream stop or timestamp rollback |
| AT-022 | Full-load acquisition | Camera, LiDAR, and all IMUs run continuously at high load for 4 hours, support simultaneous preview and recording, and produce a test report with measured performance data |
| AT-023 | External PPS timing | Valid PPS and NMEA RMC establish external UTC synchronization; loss of the source is reflected in status without falsely reporting continued lock |