Created by [ Kushant Patel], last modified by [ Mitchell Hoyt] on May 04, 2020
This term we focused on improving more on hardware aspects of sensors. Switching from 1g to 10g to provide an efficient bandwidth for all cameras, worked in collaboration with the perception team on getting our third camera setup working. Network side of sensors gave us some problems, but except for some small tweaks, we got it working. We have an automatic config script for doing a precheck on the sensor network.The script for the switching from sensor to server mode is almost ready which just needs more testing and then it can be pushed onto master for use. Being there on most test track days helped us look into more areas where sensors could pose problems.We could get the transforms working for lidars and base link setup which was pending from the last term and was on high priority.
In all, this term was perfect in terms of debugging sensors. But, there are many areas where we haven’t made any progress. We still couldn’t finish the novatel calibration task which was not on high priority but could really help the team to avoid the hard-coding of translation offsets. Also, I would like to appreciate the work of our Co-op who spent most of his days helping the division with cleanup of repo, cameras debugging, cable management and several other tasks. Collaborations are needed with perception, processing and path planning so that we can know about where sensors are giving us problems. Hardware debugging of sensors becomes an essential part because before getting into driver codes, every member of the division must know how to get the whole sensor suite up and running.\
Before the start of winter term, we had questions on whether we need more members for sensor interfacing division and as per discussion with all core members, we came to a conclusion that there’s no need for any extra members(4-5 members are enough for this division). The problem of sensor fusion is where sensors are raising new issues. This term was all about - PoEs not working, Network scripts not working, cable management not efficient, point-cloud not visible on rviz tool etc. We can’t ignore these minute details in order to ensure a smooth transition between hardware and software. Covid-19 definitely made things worse, but now that the auto-drive has been pushed to October 2020, we can improve in terms of above mentioned areas. The transparency and one-on-one between Directors , managers and core members helped us make better resolutions for blocked issues lying in our path and I hope this convention continues onward.\
+——–+—-+———-+—————+———–+—————+ | Task | Cr | Blockers | How we are | tasks to | Action | | | it | | going to fix | do | items/notes | | | ic | | this | remotely | | | | al | | | | | | | ? | | | | | +========+====+==========+===============+===========+===============+ | POE is | Me | \ | one to one | -meeting | Talk to | | not | di | | port poe's, | with | mechanical | | workin | um | | more power | power | about 4 port | | g | | | ports | systems | poe | | and | | | | and | | | sudden | | | | mechanica | | | ly | | | | l | | | turn | | | | such that | | | off | | | | the | | | | | | | integrati | | | | | | | on | | | | | | | is more | | | | | | | smooth | | +——–+—-+———-+—————+———–+—————+ | Sensor | \ | \ | - looking | \ | one port to | | and | | | into a | | take sensor | | server | | | way to | | info and | | mode | | | have | | server info | | script | | | sensors | | on, the | | | | | on one | | script | | | | | ethernet | | switches from | | | | | port and | | recieving | | | | | have | | internet | | | | | server on | | packets vs | | | | | the other | | sensor | | | | | ethernet | | packets | | | | | - Apply the | | | | | | | correct | | \ | | | | | configura | | | | | | | tion | | | | | | | such that | | | | | | | no | | | | | | | switching | | | | | | | is | | | | | | | occuring | | | +——–+—-+———-+—————+———–+—————+ | optima | \ | \ | - Have | \ | \ | | l | | | discussio | | | | placem | | | ns | | | | ent | | | with | | | | of | | | perceptio | | | | third | | | n | | | | camera | | | on how to | | | | , | | | accomplis | | | | for an | | | h | | | | optimi | | | this | | | | zed | | | - intensity | | | | field | | | of the | | | | of | | | images | | | | view | | | from the | | | | | | | new | | | | | | | camera's | | | +——–+—-+———-+—————+———–+—————+ | Novate | \ | need | \ | \ | - going to | | l | | testing | | | test | | Calibr | | to be | | | tracks to | | ation | | done | | | test it | +——–+—-+———-+—————+———–+—————+ | networ | \ | \ | \ | \ | \ | | k | | | | | | | trips | | | | | | +——–+—-+———-+—————+———–+—————+ | Cable | \ | \ | \ | \ | - novatel | | Manage | | | | | cable | | ment | | | | | managemen | | | | | | | t, | | | | | | | meeting | | | | | | | with | | | | | | | power | | | | | | | systems | | | | | | | and | | | | | | | sensor | | | | | | | interfaci | | | | | | | ng | +——–+—-+———-+—————+———–+—————+
Document generated by Confluence on Nov 28, 2021 22:40