Research for Cognitive Interaction: Issueshttps://opensource.cit-ec.de/https://opensource.cit-ec.de/favicon.ico?14265323552019-11-14T10:52:36ZResearch for Cognitive Interaction
Redmine AMiRo-OS - Feature #596 (New): automated test scripthttps://opensource.cit-ec.de/issues/5962019-11-14T10:52:36ZThomas Schöppingtschoepp@techfak.uni-bielefeld.de
<p>Introduce a test script, that sweeps compile parameters (e.g. AMIROOS_CFG_DBG), compiles all modules with these settings and evaluates compilation results (success, failure, warning) to match the expected output.<br />Since such a thorough compilation test will probably take quite some time, this should not be executed automatically by some CI tool but only on demand.</p>
<p>Preferably, this script will be included in the ./setup.sh bash script environment.</p> AMiRo-OS - Feature #591 (New): More generic QtCreator project setuphttps://opensource.cit-ec.de/issues/5912019-03-08T12:20:12ZThomas Schöppingtschoepp@techfak.uni-bielefeld.de
<p>Recently, the AMiRo-OS QtCreator project setup script has been enhanced to use the GCC output for generating according project files. This enhancement should be ported to AMiRo-BLT and enhanced, so also AMiRo-OS (as well as any further higher level projects) can take advantage of this script.</p> AMiRo-OS - Feature #589 (New): Enhanced I/O eventshttps://opensource.cit-ec.de/issues/5892019-03-08T12:13:28ZThomas Schöppingtschoepp@techfak.uni-bielefeld.de
<p>Currently I/O interrupts result in I/O events and propagate the causing EXTI line via event flags (e.g. EXTI <a href="https://opensource.cit-ec.de/issues/3" class="issue tracker-1 status-5 priority-3 priority-lowest closed" title="fix colored output in installink target (Closed)">#3</a> propagates (1 << 3) = 0x00000008). Since EXTI lines can aggregate multiple I/O pins (e.g. PA3 and PC3) this method can result in ambiguous events, because only the EXTI line but not the actual pin is specified.</p>
<p>To solve this issue, I/O events should not propagate plain eventflags_t data, but a struct like<br /><pre>
struct {
uint32_t flags : 24; // alternatively 28
uint32_t pin : 8; // alternatively 4
};
</pre>Since there is no EXTI hardware so far which provides more than 24 lines this solution is feasible. Alternatively the ratio can be set to 28:4 assuming that there will not be more than 2^4-1 pins aggregated in a single EXTI line. The -1 is important here, since the value of <code>pin</code> = 0 must be reserved to indicate ambiguity. This was the new method is compatible to plain eventflags_t and there might be cases where the exact pin can not be determined.<br />The determination can be realized within the ISR, which is called by the EXTI driver.</p> Cognitive Interaction Toolkit - Feature #584 (New): citkat: Indicate external links using an icon...https://opensource.cit-ec.de/issues/5842018-10-16T08:38:22ZJan Moringenjmoringe@techfak.uni-bielefeld.de
<p>Links to project homepages and bug trackers, for example.</p> hlrc - Bug #581 (New): flobi lookat does not work anymore due to hardcoded floka prefixhttps://opensource.cit-ec.de/issues/5812018-08-15T09:20:26ZGuillaume Walckgwalck@techfak.uni-bielefeld.de
<p>I am using flobi, and we get this error now :<br /><pre>
2018-08-15 10:39:20,004 hlrc_client.Middleware ERROR > publish_lookat_target timed out waiting for result or soemthing else went wrong
and in the log of hlrc_server :
[ WARN] [1534322517.517683315]: "floka_BASE_LINK" passed to lookupTransform argument target_frame does not exist
</pre></p>
<p>I tracked down the error to a recent change with a hardcoded prefix.</p>
<p><a class="external" href="https://opensource.cit-ec.de/projects/hlrc/repository/diff/server/include/ROS/LookatCallbackWrapperROS.h?utf8=%E2%9C%93&rev=01ff84641a5a769064e8656fcfd4d48f69a24d18&rev_to=60b91de0f20bdd7fac9fb448b2fad1079a7b9891">https://opensource.cit-ec.de/projects/hlrc/repository/diff/server/include/ROS/LookatCallbackWrapperROS.h?utf8=%E2%9C%93&rev=01ff84641a5a769064e8656fcfd4d48f69a24d18&rev_to=60b91de0f20bdd7fac9fb448b2fad1079a7b9891</a></p>
<p>Maybe the prefix could be taken from the scope name (unless you want to make that cleaner) that is passed to the arguments anyway<br /><pre>
hlrc_server ROS ROS flobi flobi pulse
</pre></p> Image Component Library - Bug #528 (New): camera render in scene if pos is 0,0,0https://opensource.cit-ec.de/issues/5282017-11-09T11:22:13ZGuillaume Walckgwalck@techfak.uni-bielefeld.de
<p>If the camera extrinsic is identity (pos is 0,0,0), rendering in the scene creates errors</p>
<pre>
[/media/local/jenkins/jobs/icl-nivision-trunk-ci-deploy-lsp-famula-nightly/workspace/ICLGeom/src/ICLGeom/Scene.cpp:prepareForRendering,line: 149] WARNING: error visualizsing camera: no intersection -> plane normal is perdendicular to view-ray direction
</pre>
<p>Looking deeper, here is maybe the explanation :</p>
<p>When creating the vertices of the pyramid/frustrum of the camera, the 4 base corners are defined as the intersection of the image plane with the "viewRays" of the 4 image corner ( 0,0 / 0,479 / 639,479 / 639,0). However, this view rays never intersec the plane if they are in the plane which happens when position of the camera is zero. <br />See Scene.cpp:142 and Scene.cpp:145 as well as Camera.cpp:720 and Camera.cpp:770</p>
<p>there might me more to this issue, for instance why can the camera center and the image plane be at the same place, there should be focal length in between at all time. This comes maybe from the depth image model which considers the plane to be at zero Z in order to mesure depth from there.... then maybe the position of the camera center should always substract the focal length...</p> Cognitive Interaction Toolkit - Bug #525 (New): Maven Template uses default profilehttps://opensource.cit-ec.de/issues/5252017-11-02T07:38:33ZLeroy Rügemerlruegeme@techfak.uni-bielefeld.de
<p>Maven builds use the users .m2 settings file maven should create and use special settings for each.</p>
<p>To build multiple distributions on the same machine i have to switch the default profile to not pull from the wrong distribution</p> Image Component Library - Bug #499 (New): Does not build with C++98https://opensource.cit-ec.de/issues/4992017-02-20T14:54:57ZTimo Korthalstkorthals@cit-ec.uni-bielefeld.de
<p>The library does not build with the standard settings of gcc, so why not directly setting the standard to <code>c++0x</code>?</p>
<pre>
/opt/repositories/ICL/ICLCV/src/ICLCV/LensUndistortionCalibrator.cpp:177:70: error: in C++98 ‘s’ must be initialized by constructor, not by ‘{...}’
CvSize s = { m_data->imageSize.width, m_data->imageSize.height };
</pre> IPAACA - Bug #496 (New): RSB calls should never blockhttps://opensource.cit-ec.de/issues/4962017-01-26T14:17:42ZHendrik Buschmeier
<p>Calling a non-existen remote Server, e.g., results in the process blocking indefinitely.</p> IPAACA - Bug #495 (New): Updates get applied even if revision is outdatedhttps://opensource.cit-ec.de/issues/4952017-01-26T14:16:31ZHendrik Buschmeier
<p>Currently IU updates are applied without checking the local revision of the IU in the remote InputBuffer.</p>
<p>This can, e.g., happen when multiple updates are issued in short succession, the first of which triggers a resendRequest, which returns a revision that is newer than multiple in-between updates.</p>
<pre>
A: IU1.publish – r1
B: startup
A: IU1.update – r2
B: IU1.applyUpdate -> fail
B: requestResend(IU1)
A: IU1.update – r3
A: IU1.update – r4
A: reply requestResend(U1) – r4
B: IU1.added – r4
B: IU1.applyUpdate – r3
B: IU1.applyUpdate – r4
</pre> Finite State Machine Based Testing - Bug #492 (New): Installation fails on a fresh Ubuntu 14.04.4https://opensource.cit-ec.de/issues/4922016-11-30T21:28:32ZSebastian Wredeswrede@techfak.uni-bielefeld.de
<pre>
Installed /home/cosima/citk/systems/cogimon-minimal-trusty-nightly/lib/python2.7/site-packages/enum34-1.1.6-py2.7.egg
error: Installed distribution setuptools 3.3 conflicts with requirement setuptools>=11.3
Build step 'Execute shell' marked build as failure
</pre>
<p>This can be reproduced on a clean Ubuntu 14.04.4 image using the <code>cogimon-minimal-trusty-nightly</code> distribution.</p> IPAACA - Feature #488 (New): ipaaca_pb2 should lie inside ipaaca package and in the PYTHONPATHhttps://opensource.cit-ec.de/issues/4882016-09-12T11:24:11ZHendrik Buschmeier
<p>ipaaca is currently not self-contained as the protobuf generated ipaaca_pb2.py file resides outside the ipaaca package. Although it is sufficient to just copy it into the package, the including files should import it as ipaaca.ipaaca_pb2 instead of just ipaaca_pb2.</p>
What needs to be done:
<ul>
<li>Change build process so that it puts ipaaca_pb2 in the right place.</li>
<li>Change ipaaca sub-modules to import ipaaca.ipaaca_pb2 instead of just ipaaca_pb2.</li>
</ul> IPAACA - Feature #467 (New): v3: Linked IU events / Batch updates on the IU [event] levelhttps://opensource.cit-ec.de/issues/4672016-02-16T13:45:37ZRamin Yaghoubzadehryaghoubzadeh@uni-bielefeld.de
<p>Similarly to the batch payload updates introduced in v2, v3 ought to provide a means of coordinating several IU events in a single indivisible transaction.</p>
<p>A possible, and important, use case would be the publication of linked sets of IUs in a single shot.</p>
<p>This will most certainly involve support on the protocol level (note that <a href="https://opensource.cit-ec.de/issues/462" class="issue tracker-2 status-2 priority-4 priority-default child parent" title="v3: Structured metadata (In Progress)">#462</a> will entail a new protocol version, anyway).</p> IPAACA - Feature #464 (New): v3: Sender tokenshttps://opensource.cit-ec.de/issues/4642016-02-16T13:23:28ZRamin Yaghoubzadehryaghoubzadeh@uni-bielefeld.de
<p>In lieu of a complete authentication / encryption layer, v3 will include a simple token-based filtering mechanism, probably defaulting to a username-derived personal token. Components can override their sending token, and also the default filter of allowing reception of IU events with the user's token only.<br />In so doing, basic concurrent usage of different users / demos is made feasible – while not excluding the possibility to wantonly modify a closed system when required.</p> Cognitive Interaction Toolkit - Feature #287 (New): Read branches out of git to reduce versioning...https://opensource.cit-ec.de/issues/2872014-09-18T13:17:51ZMarian Pohlingmpohling@cit-ec.uni-bielefeld.de
<p>Would it be possible to read the "branches"/"tags" direktly out the git repo instead of listing all branches explicit in the recipe file again.</p>