picoScan100: 3D Point Clouds with an External Encoder

1 - Concept and Approach Selection

Two distinct approaches exist for capturing and assembling a 3D point cloud from a picoScan150 mounted on a motorised tilt axis:

The picoScan150 measures distance and intensity in a single plane. Mounting it on a motorised tilt axis sweeps the scan plane through 3D space. If an incremental encoder is attached to the axis, the rotation angle at the moment of each scan is known precisely. Each 2D point can then be rotated into a common 3D coordinate frame, and the accumulated points from a full sweep form a dense 3D point cloud.

The sensor supports encoder input directly and can tag each UDP scan packet with the encoder tick at capture time. This frame-accurate synchronisation avoids timing jitter between scan data and position data even when both streams are received over the same network connection.

2 - Hardware Setup

The hardware setup below describes one typical example: a sensor mounted on a motorised tilt axis. The same approach applies to other motion types - for example, a linear translation stage or a rotation around a different axis. What matters is that an incremental encoder is attached to the moving axis and wired to the sensor's encoder input connector so that each scan frame can be tagged with a position value.

Required components for this example:

• A picoScan150 with encoder input and compact-format UDP output
• An incremental encoder wired to the sensor's encoder input connector (e.g. SICK DFS60 series)
• A motorised axis moving the sensor (tilt, rotation, or linear translation)
• Network connection to the host running the application

2.1  Example axis geometry (tilt axis)

In this example, the rotation axis runs along the world Y–axis and the scan plane is horizontal (XZ–plane) at the reference position. As the axis tilts, the tilt angle θ increases from the reference. For a different axis orientation or a linear stage, the coordinate transform in §4 must be adapted accordingly.

The encoder produces incremental ticks. A full revolution corresponds to a fixed number of ticks, which depends on the encoder model. Reset the encoder counter at the desired reference position to establish tick 0 as the angular origin.

2.2  Encoder output ports

3 - Frame-Accurate Encoder Synchronisation

Both UDP streams - scan data and encoder data - carry a frameSequenceNumber field. This number increments with every scan frame and is embedded in both the scan packet and the matching encoder packet. By building a lookup table that maps each frame sequence number to its encoder tick, the application can retrieve the exact axis angle for any scan frame even when the two UDP streams arrive slightly out of order.

// On every encoder packet:
encoder_map[packet.frameSequenceNumber] = packet.tickCounter

// On every scan frame:
tick = encoder_map.lookup(frame.frameSequenceNumber)
        // or use the most recent tick if the entry is missing

Keep the lookup table bounded to avoid unbounded memory growth during long acquisition sessions.

4 - Coordinate Transform

The 2D scan returns Cartesian points (x, y) in the sensor plane. To convert these to 3D world coordinates, rotate each point around the Y–axis by the tilt angle θ at the time of capture.

The angle is computed from the encoder tick relative to the reference position (tick 0 after a counter reset):

theta = (tick / ticks_per_revolution) * 2 * PI   // radians

The Y–axis rotation transforms each scan point (x_scan, y_scan) to world 3D:

x_world =  x_scan * cos(theta)
y_world =  y_scan                  // lateral axis unchanged
z_world =  x_scan * sin(theta)

5 - End-to-End Pseudocode

// ── Setup ──────────────────────────────────────────────────────────────

// Reset encoder counter to establish the angular reference (tick 0)
http_post("/api/methods/mResetEncoderIncrement")

// ── Data Collection ────────────────────────────────────────────────────

encoder_map = {}                // frame_seq_num -> tick
cloud_pts   = []                // accumulated 3D points

// Encoder packet callback
on_encoder_packet(pkt):
    encoder_map[pkt.frameSequenceNumber] = pkt.tickCounter

// Scan frame callback
on_scan_frame(frame, points_2d):
    tick  = encoder_map.get(frame.prevSegmentFrameSeqNum, last_known_tick)
    theta = (tick / TICKS_PER_REV) * 2 * PI

    for (x, y, intensity) in points_2d:
        cloud_pts.append({
            x: x * cos(theta),
            y: y,
            z: x * sin(theta),
            intensity: intensity
        })

// Open UDP sockets and start receiving
udp_listen(port=2115, callback=on_scan_frame)
udp_listen(port=7504, callback=on_encoder_packet)
run_until_interrupted()

// Save accumulated points when acquisition is complete
save_pcd(cloud_pts, "cloud.pcd")

6 - On-Device Scan Merger: Overview

Firmware V2.3.0 introduces a scan merger that runs entirely on the device. While the sensor rotates, it combines consecutive 2D scan profiles into a single 3D point cloud and saves the result as a .pcd file in device storage. The host triggers the recording, polls for completion, and downloads the file with one HTTP GET request. No UDP parsing or coordinate transform code is required on the host.

The angle source is configurable. This section focuses on the encoder source: the same incremental encoder wired to the sensor is used, but the tick-to-angle conversion runs entirely on the device instead of the host. The built-in IMU is also supported as an angle source when no encoder is available.

Workflow summary:

1. Configure the scan merger parameters and persist them to EEPROM.
2. Reset the on-device point cloud view to clear any previous recording.
3. Start the merger; it runs until the host sends a stop command.
4. Poll MergeRunning until the recording duration is reached, then call StopScanMerge.
5. Poll MergeSaving and SaveProgress until the PCD file is written (100%).
6. Download the file from GET /cloud.pcd.

7 - Scan Merger Configuration and Authentication

7.1  Configuration parameters

8 - Merger Workflow and State Reference

8.1  State fields

8.2  Recording and download steps

9 - On-Device Merger: End-to-End Pseudocode

// write_api() represents an authenticated POST to /api/<endpoint>.
// For authentication implementation see:
//   https://github.com/SICKAG/sick_perception_sdk
//   https://github.com/SICKAG/ScanSegmentAPI

// ── 1. Configure scan merger ───────────────────────────────────────────
write_api("ScanMergerEnabled",  { "ScanMergerEnabled": true })
write_api("ScanMergerSource",   { "ScanMergerSource":  "ENCODER" }) // or "IMU"
write_api("ScanMergeTrigger",   { "ScanMergeTrigger":  0 })        // 0 = METHOD
write_api("RotationOffset",     { "RotationOffset": {"x":0,"y":0,"z":0} })

// ── 2. Reset view ─────────────────────────────────────────────────────
write_api("methods/ResetScanMergeView", {})

// ── 3. Start recording ────────────────────────────────────────────────
write_api("methods/StartScanMerge", {})

start_time = now()
while True:
    sleep(2000 ms)
    running = http_get("/api/MergeRunning").data.MergeRunning
    if now() - start_time >= DURATION_SECONDS:
        break

// ── 4. Stop recording ─────────────────────────────────────────────────
write_api("methods/StopScanMerge", {})

// ── 5. Wait for on-device save ────────────────────────────────────────
deadline = now() + 120 s
while now() < deadline:
    sleep(2000 ms)
    saving   = http_get("/api/MergeSaving").data.MergeSaving
    progress = http_get("/api/SaveProgress").data.SaveProgress
    if not saving and progress == 100:
        break

// ── 6. Download PCD ───────────────────────────────────────────────────
response = http_get("http://<device-ip>/cloud.pcd")   // no auth required
write_file("cloud.pcd", response.body)
print("Saved", len(response.body) / 1024, "KB")