<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Starling 2 Max – VIO/Odometry Reliability Issues & Purchase of a Second Unit with Validated Software]]></title><description><![CDATA[<h2>Support Information</h2>
<p dir="auto"><strong>Hardware:</strong></p>
<ul>
<li><strong>Aircraft:</strong> ModalAI Starling 2 Max (C29)</li>
<li><strong>SKU:</strong> <code>MRB-D0012-4-V4-C29-T8-M36</code></li>
<li><strong>VOXL platform:</strong> VOXL2 M0054</li>
<li><strong>Tracking cameras:</strong> <code>tracking_front</code> and <code>tracking_down</code> (AR0144)</li>
<li><strong>IMU:</strong> ICM42688</li>
</ul>
<p dir="auto"><strong>Software version:</strong></p>
<ul>
<li><strong>System image:</strong> <code>1.8.08-M0054-14.1a-perf</code></li>
<li><strong>VOXL suite:</strong> <code>1.6.4~beta5</code></li>
<li><strong>OpenVINS:</strong> <code>voxl-open-vins-server 0.6.1</code></li>
</ul>
<p dir="auto">These are the software versions recorded during our investigation.</p>
<p dir="auto"><strong>Configuration / services:</strong></p>
<ul>
<li>Two-camera OpenVINS configuration using <code>tracking_front</code> and <code>tracking_down</code>.</li>
<li>Official configuration preset: <code>starling2_2cam</code>.</li>
<li>IMU service: <code>voxl-imu-server</code>.</li>
<li>VIO service: <code>voxl-open-vins-server</code>.</li>
<li>OpenVINS was recently restored using <code>voxl-configure-open-vins starling2_2cam</code>.</li>
<li>Previous diagnostic tests included changes to online camera-extrinsic calibration settings.</li>
<li>A startup-order issue between OpenVINS and the IMU service was identified.</li>
</ul>
<p dir="auto"><strong>What I am trying to do:</strong></p>
<p dir="auto">We are developing autonomous indoor navigation using the Starling 2 Max and require reliable, repeatable metric odometry.</p>
<p dir="auto">Our environment is an ordinary indoor workshop containing shelving, storage racks, tables, and workbenches. There is nothing particularly unusual about the environment.</p>
<p dir="auto">We also intend to purchase a <strong>second identical Starling 2 Max</strong>, while retaining our existing aircraft.</p>
<p dir="auto">We would like the new unit to arrive with the <strong>most recent complete software configuration officially validated for this specific aircraft</strong>, including any available improvements relevant to OpenVINS reliability.</p>
<p dir="auto"><strong>What actually happens:</strong></p>
<p dir="auto">OpenVINS produces substantially different metric-scale estimates between diagnostic sessions.</p>
<p dir="auto">During approximately 1.4-meter vertical movements, we measured estimated displacements of:</p>
<ul>
<li><strong>1.41 m:</strong> approximately correct.</li>
<li><strong>0.32 m:</strong> approximately 23% of the physical displacement.</li>
<li><strong>0.46 m:</strong> approximately 33% of the physical displacement.</li>
</ul>
<p dir="auto">Horizontal odometry also exhibited inconsistent scale.</p>
<p dir="auto">Additional observations:</p>
<ul>
<li>OpenVINS occasionally initializes before the IMU service is ready.</li>
<li>IMU propagation-window configuration differs depending on startup conditions.</li>
<li>Timestamp regressions of approximately 33.3 ms have occurred.</li>
<li><code>BAD_TIMESTAMP</code> errors have triggered OpenVINS automatic resets.</li>
<li>Visual-tracking quality and trajectory closure errors vary considerably between test sessions.</li>
</ul>
<p dir="auto">The root cause of the metric-scale inconsistency remains undetermined.</p>
<p dir="auto"><strong>Steps to reproduce:</strong></p>
<ol>
<li>Start the aircraft and initialize the two-camera OpenVINS system.</li>
<li>Confirm that the IMU and tracking cameras are operating.</li>
<li>Manually carry the aircraft through an approximately 1.4-meter vertical movement.</li>
<li>Execute an approximately 2 m × 2 m horizontal square trajectory.</li>
<li>Return toward the starting position.</li>
<li>Compare OpenVINS displacement against the approximate physical movement.</li>
<li>Repeat under another startup or initialization condition.</li>
</ol>
<p dir="auto">These were manually executed diagnostic movements without motor actuation. Some sessions involved configuration changes and are not identical controlled repetitions.</p>
<p dir="auto"><strong>Logs / pictures:</strong></p>
<p dir="auto">No files are attached to this post.</p>
<p dir="auto">Relevant errors observed during our investigation:</p>
<p dir="auto"><code>WARNING: could not read IMU pipe info JSON for 'imu_apps'</code></p>
<p dir="auto"><code>[HEALTH] ERROR: Sensor measurements with bad timestamps</code></p>
<p dir="auto"><code>AUTO-RESET RECOMMENDED Error code(s) detected 0x8000</code></p>
<p dir="auto">The measured results and technical observations are summarized in the following report.</p>
<hr />
<h1>Detailed Technical Context</h1>
<h2>1. Background and purpose</h2>
<p dir="auto">Hello ModalAI Engineering Team,</p>
<p dir="auto">We currently own a <strong>Starling 2 Max (C29)</strong>, which we use for autonomous indoor robotics development.</p>
<p dir="auto">We are planning to purchase a <strong>second identical Starling 2 Max</strong> and would appreciate your engineering recommendations regarding the most appropriate software configuration for the new aircraft.</p>
<p dir="auto"><strong>We intend to keep our existing drone. We are not requesting a replacement or exchange.</strong></p>
<p dir="auto">Our objective is to continue using the existing aircraft for development while acquiring an additional unit with the most recent, officially validated software available for this particular hardware.</p>
<p dir="auto">The reason for our inquiry is that we have encountered significant inconsistencies in <strong>OpenVINS metric-scale estimation</strong>.</p>
<p dir="auto">We have investigated multiple possible contributing factors, including initialization behavior, camera–IMU synchronization, visual-tracking quality, calibration settings, and timestamp ordering.</p>
<p dir="auto">We have not conclusively determined the cause.</p>
<p dir="auto">However, several recent ModalAI forum discussions and OpenVINS software developments appear relevant to our observations.</p>
<p dir="auto">We would appreciate your engineering assessment and clarification on whether the newer software configurations address any of these behaviors.</p>
<h2>2. Operating environment</h2>
<p dir="auto">Our drone operates in an ordinary indoor workshop.</p>
<p dir="auto">The environment contains shelves, storage racks, tables, workbenches, and other typical stationary workshop objects.</p>
<p dir="auto">It is not a specialized motion-capture laboratory or an unusual visual environment.</p>
<p dir="auto">We understand that visual-inertial odometry performance depends on factors such as lighting, feature availability, camera exposure, motion, synchronization, and calibration.</p>
<p dir="auto">We are not claiming that all environmental or visual-tracking factors have been eliminated.</p>
<p dir="auto">However, our primary concern is not ordinary odometry drift.</p>
<p dir="auto"><strong>Our concern is the significant variation in metric scale between test sessions.</strong></p>
<h2>3. Observed metric-scale inconsistency</h2>
<p dir="auto">We performed several manually executed experiments with the aircraft's motors inactive.</p>
<p dir="auto">A typical test involved an approximately measured vertical ascent of 1.4 meters, followed by a horizontal square trajectory with approximately 2-meter sides.</p>
<p dir="auto">The objective was to evaluate how accurately OpenVINS represented the physical movement.</p>
<p dir="auto">Representative vertical measurements included:</p>
<table class="table table-bordered table-striped">
<thead>
<tr>
<th>Test</th>
<th style="text-align:right">Approximate physical movement</th>
<th style="text-align:right">OpenVINS displacement</th>
<th style="text-align:right">Scale</th>
</tr>
</thead>
<tbody>
<tr>
<td>A</td>
<td style="text-align:right">1.40 m</td>
<td style="text-align:right">1.41 m</td>
<td style="text-align:right">1.01×</td>
</tr>
<tr>
<td>B</td>
<td style="text-align:right">1.40 m</td>
<td style="text-align:right">0.32 m</td>
<td style="text-align:right">0.23×</td>
</tr>
<tr>
<td>C</td>
<td style="text-align:right">1.40 m</td>
<td style="text-align:right">0.46 m</td>
<td style="text-align:right">0.33×</td>
</tr>
</tbody>
</table>
<p dir="auto">We also observed inconsistent horizontal-scale estimates during approximately 2-meter movements.</p>
<p dir="auto">Some sessions produced estimates relatively close to the physical trajectory, while others substantially underestimated displacement.</p>
<p dir="auto">We recognize that these were hand-carried tests rather than motion-capture measurements.</p>
<p dir="auto">They also represent different diagnostic sessions, some of which involved changes to the software configuration.</p>
<p dir="auto">Nevertheless, the observed differences are large enough to raise concerns about initialization consistency or another aspect of the VIO estimation process.</p>
<p dir="auto"><strong>We do not know whether the cause is software initialization, camera–IMU calibration, timestamp handling, visual tracking, or another factor.</strong></p>
<p dir="auto">This is the central technical question we would like ModalAI engineering to address.</p>
<h2>4. IMU and OpenVINS startup behavior</h2>
<p dir="auto">During our investigation, we identified an issue involving the startup order of OpenVINS and the IMU service.</p>
<p dir="auto">On some starts, OpenVINS initialized before the IMU service had completed its initialization.</p>
<p dir="auto">The journal reported:</p>
<p dir="auto"><code>WARNING: could not read IMU pipe info JSON for 'imu_apps'</code></p>
<p dir="auto">In those conditions, OpenVINS used:</p>
<p dir="auto"><code>prop_window: 0.075</code></p>
<p dir="auto">When we ensured that the IMU service was fully ready before restarting OpenVINS, the system correctly identified the ICM42688 and used:</p>
<p dir="auto"><code>prop_window: 0.100</code></p>
<p dir="auto">We reproduced this behavior multiple times.</p>
<p dir="auto">However, correcting the startup sequence did not consistently eliminate the metric-scale inconsistency.</p>
<p dir="auto">We would appreciate clarification on whether this startup behavior is known in OpenVINS 0.6.1 and whether more recent software releases improve service initialization or IMU detection.</p>
<p dir="auto">We are not suggesting that the propagation-window difference has been proven responsible for the scale problem.</p>
<h2>5. Timestamp-ordering anomalies</h2>
<p dir="auto">We also identified an independent timing anomaly during OpenVINS operation.</p>
<p dir="auto">In one diagnostic session, OpenVINS reported an out-of-order VIO pose with a timestamp regression of approximately <strong>33.3 milliseconds</strong>.</p>
<p dir="auto">This corresponds approximately to one camera-frame period at 30 Hz.</p>
<p dir="auto">The journal subsequently reported:</p>
<p dir="auto"><code>[HEALTH] ERROR: Sensor measurements with bad timestamps</code></p>
<p dir="auto">Followed by:</p>
<p dir="auto"><code>AUTO-RESET RECOMMENDED Error code(s) detected 0x8000</code></p>
<p dir="auto">OpenVINS then performed an automatic hard reset.</p>
<p dir="auto">Our inspection of the corresponding raw camera and IMU timestamp streams did not reveal the same backward timestamp regression.</p>
<p dir="auto">This suggests the possibility of an ordering problem within the subsequent VIO processing or publication chain, although we have not isolated its exact origin.</p>
<p dir="auto">Importantly, we also observed incorrect metric-scale estimates during sessions without automatic resets.</p>
<p dir="auto">Therefore, we consider the timestamp anomaly a separately documented problem, <strong>not a confirmed explanation for the scale inconsistency</strong>.</p>
<p dir="auto">We would appreciate knowing whether this behavior is recognized in OpenVINS 0.6.1 and whether it has been addressed in later versions.</p>
<h2>6. Visual-tracking quality</h2>
<p dir="auto">We also observed significant variation in the visual-tracking quality reported by OpenVINS.</p>
<p dir="auto">Two diagnostic sessions conducted on October 1 produced the following measurements:</p>
<table class="table table-bordered table-striped">
<thead>
<tr>
<th>Measurement</th>
<th style="text-align:right">Session 14:16</th>
<th style="text-align:right">Session 14:24</th>
</tr>
</thead>
<tbody>
<tr>
<td>Estimated trajectory length</td>
<td style="text-align:right">11.81 m</td>
<td style="text-align:right">14.19 m</td>
</tr>
<tr>
<td>Final position closure error</td>
<td style="text-align:right">38.5 cm</td>
<td style="text-align:right">93.9 cm</td>
</tr>
<tr>
<td>Closure error relative to estimated trajectory</td>
<td style="text-align:right">3.3%</td>
<td style="text-align:right">6.6%</td>
</tr>
<tr>
<td>Median VIO quality</td>
<td style="text-align:right">81</td>
<td style="text-align:right">30</td>
</tr>
</tbody>
</table>
<p dir="auto">These were independent hand-carried trajectories rather than identical controlled repetitions.</p>
<p dir="auto">Nevertheless, they demonstrate substantial differences in reported tracking quality and final position consistency between sessions.</p>
<p dir="auto">We also investigated OpenVINS camera-calibration settings, including temporarily disabling online camera-extrinsic calibration.</p>
<p dir="auto">That experiment did not provide a consistent solution.</p>
<p dir="auto">More recently, we restored the OpenVINS configuration using the official:</p>
<p dir="auto"><code>voxl-configure-open-vins starling2_2cam</code></p>
<p dir="auto">We have not established whether the latest configuration reset changes the behavior.</p>
<h2>7. Related ModalAI forum findings</h2>
<p dir="auto">Our investigation led us to several ModalAI discussions and more recent software developments.</p>
<p dir="auto">We believe these findings may be relevant, although we have not demonstrated a causal relationship with our metric-scale problem.</p>
<h3>A. OpenVINS initialization behavior on the same hardware</h3>
<p dir="auto">We reviewed the forum topic:</p>
<p dir="auto"><strong>“voxl-mapper produces no map; VIO output rate halves to 15 Hz and does not recover until OpenVINS restart.”</strong></p>
<p dir="auto">This discussion involved the same Starling 2 Max C29 SKU and OpenVINS 0.6.1.</p>
<p dir="auto">ModalAI engineering discussed an initialization-related mechanism involving static-jerk detection and camera-frame processing rates.</p>
<p dir="auto">The discussion also clarified that apparently healthy VIO output can occur while the system is publishing blank VIO packets rather than actively estimating visual-inertial movement.</p>
<p dir="auto">We found this particularly interesting because our estimated metric scale has varied substantially between diagnostic sessions.</p>
<p dir="auto">We have not independently verified the specific 15 Hz behavior described in that thread.</p>
<p dir="auto">However, it raises the following question:</p>
<p dir="auto"><strong>Could differences in OpenVINS initialization or estimator activation contribute to inconsistent metric-scale estimation between sessions?</strong></p>
<h3>B. AR0144 synchronization and updated timestamp handling</h3>
<p dir="auto">We also reviewed the forum discussion titled:</p>
<p dir="auto"><strong>“OpenVINS”</strong></p>
<p dir="auto">In this discussion, ModalAI referenced updated OpenVINS code and a <code>synced-timestamp</code> branch for synchronized AR0144 tracking cameras.</p>
<p dir="auto">Our aircraft uses AR0144 tracking cameras, and we have observed timestamp-ordering anomalies.</p>
<p dir="auto">We would appreciate clarification on whether those changes are applicable to our hardware and software configuration.</p>
<p dir="auto">We are particularly interested in whether OpenVINS 0.6.1 has known limitations involving synchronized camera processing that have been addressed in later releases.</p>
<h3>C. OpenVINS 0.7.8 and SDK 1.8</h3>
<p dir="auto">We also reviewed the SDK 1.8 release notes, including the changes introduced in:</p>
<p dir="auto"><code>voxl-open-vins-server 0.7.8</code></p>
<p dir="auto">We understand that the newer implementation includes improvements involving:</p>
<ul>
<li>Asynchronous multicamera processing.</li>
<li>Timestamp-ordered camera-frame handling.</li>
<li>Visual-inertial initialization.</li>
<li>Calibration and diagnostic functionality.</li>
</ul>
<p dir="auto">These changes appear relevant to two areas of our investigation:</p>
<p dir="auto"><strong>Initialization consistency and timestamp ordering.</strong></p>
<p dir="auto">We are not assuming that OpenVINS 0.7.8 resolves our specific observations.</p>
<p dir="auto">However, we would appreciate confirmation of whether these improvements address known limitations of the earlier implementation and whether the newer version is officially supported and validated for Starling 2 Max C29.</p>
<h2>8. Request regarding the purchase of another Starling 2 Max</h2>
<p dir="auto">This is the main reason for contacting ModalAI.</p>
<p dir="auto">We intend to purchase a <strong>second Starling 2 Max with the same hardware configuration as our current aircraft</strong>.</p>
<p dir="auto">We will retain our existing drone.</p>
<p dir="auto">Our priority is to receive the new aircraft with the most reliable and recent <strong>complete software configuration that ModalAI has officially validated for this specific model</strong>.</p>
<p dir="auto">We do not necessarily want the highest available software version.</p>
<p dir="auto">We want the most appropriate manufacturer-supported combination of SDK, OpenVINS, camera services, IMU services, PX4 integration, and supporting components.</p>
<p dir="auto">We noticed that the published compatibility information identifies SDK 1.6.4 as a flight-tested release for Starling 2 Max D0012, while newer SDK releases contain significant OpenVINS developments.</p>
<p dir="auto">We would appreciate direct clarification on the following questions.</p>
<h3>1. Recommended software baseline</h3>
<p dir="auto"><strong>What is the newest complete VOXL SDK release currently recommended and officially flight-tested by ModalAI for Starling 2 Max C29?</strong></p>
<p dir="auto">Is there a newer validated configuration than the software installed on our current aircraft?</p>
<h3>2. Factory delivery with validated software</h3>
<p dir="auto"><strong>Can ModalAI deliver a newly purchased Starling 2 Max with that recommended software already installed, configured, calibrated, and verified?</strong></p>
<p dir="auto">We would prefer a complete manufacturer-supported configuration rather than independently upgrading individual services.</p>
<h3>3. Metric-scale consistency</h3>
<p dir="auto"><strong>Is ModalAI aware of any initialization, observability, calibration, or estimation behavior in OpenVINS 0.6.1 that could explain substantial variations in metric scale between sessions?</strong></p>
<p dir="auto">Have any related issues been corrected or improved in newer releases?</p>
<h3>4. Timestamp and multicamera synchronization</h3>
<p dir="auto"><strong>Are the changes associated with the <code>synced-timestamp</code> branch or the newer asynchronous multicamera architecture relevant to our AR0144 configuration and the approximately 33.3 ms timestamp regressions we observed?</strong></p>
<p dir="auto">Is a corrected implementation included in a validated software release for this aircraft?</p>
<h3>5. OpenVINS 0.7.8 compatibility</h3>
<p dir="auto"><strong>Is OpenVINS 0.7.8, or a later equivalent, officially supported and flight-tested on Starling 2 Max C29?</strong></p>
<p dir="auto">If not, what version does ModalAI recommend?</p>
<h3>6. Factory calibration and verification</h3>
<p dir="auto"><strong>Can the new aircraft be delivered with its original camera intrinsics, camera–IMU extrinsics, IMU calibration, and confirmation of the exact software versions installed?</strong></p>
<p dir="auto">Does ModalAI perform VIO initialization and basic odometry verification before shipment?</p>
<h3>7. Recommended configuration for our application</h3>
<p dir="auto"><strong>Given the hardware, software versions, and observed behavior described above, what complete software configuration would ModalAI engineering recommend for reliable autonomous indoor navigation?</strong></p>
<p dir="auto">We are particularly interested in avoiding known initialization and synchronization limitations.</p>
<h2>9. Final objective</h2>
<p dir="auto">Our objective is to establish a reliable, repeatable, manufacturer-supported VIO baseline.</p>
<p dir="auto">We have observed significant metric-scale inconsistency, along with separate initialization, timestamp-ordering, and tracking-quality anomalies.</p>
<p dir="auto">We have also identified software developments that appear relevant, but we have not demonstrated that they resolve our particular problem.</p>
<p dir="auto">We are not concluding that our existing aircraft has defective hardware or that a specific software bug has been proven responsible.</p>
<p dir="auto"><strong>We would like ModalAI engineering to clarify which of these behaviors are known, whether newer validated software addresses them, and what configuration should be installed on a newly purchased Starling 2 Max.</strong></p>
<p dir="auto">We intend to keep our current aircraft and purchase a second identical unit.</p>
<p dir="auto">Ideally, the new aircraft would arrive with the most recent officially validated software configuration, including any confirmed improvements relevant to OpenVINS initialization, camera synchronization, and metric-scale reliability.</p>
<p dir="auto">We would appreciate a clear technical recommendation and confirmation of whether this factory configuration can be arranged as part of the purchase.</p>
<p dir="auto">Thank you for your time and assistance.</p>
<p dir="auto">Best regards,</p>
<p dir="auto">Jorge Padilla</p>
]]></description><link>https://forum.modalai.com/topic/5550/starling-2-max-vio-odometry-reliability-issues-purchase-of-a-second-unit-with-validated-software</link><generator>RSS for Node</generator><lastBuildDate>Tue, 17 Nov 2026 01:50:55 GMT</lastBuildDate><atom:link href="https://forum.modalai.com/topic/5550.rss" rel="self" type="application/rss+xml"/><pubDate>Sat, 14 Nov 2026 22:51:53 GMT</pubDate><ttl>60</ttl></channel></rss>