The week of 26 October to 1 November 2026
- Recap
- 2 of 3
- Date
- Reading
- 2 min read
This recap is a week late. That is more or less the story of the week: the robot spent three days switched off, and I spent three days looking for the reason.
What happened
- One error, twenty hours powered. On Tuesday around 3 a.m. the program stopped on a timing error from the daemon, in the middle of a 0.35-second glance. It only knew how to catch network errors. The robot stayed up, motors powered, for about twenty hours. Since then, a failed gesture is just a warning, and any exit of the program puts the robot to bed. Read the post.
- What the official app costs. With the camera open, the processor load climbs to 7 on 4 cores. Four minutes after closing it, it is back to 2.4. Read the post.
- Three silent days. From Thursday evening to Sunday, nothing: no network, not a single LED on the head board. I thought the board was dead. It was the power cable from the base, pulled out of its connector for lack of slack. Lesson, now first on the list: before you condemn a board, follow the cable from the base to the head. Read the post.
- The battery becomes readable. The nine motors report the voltage. An “energy” tile on the slate now says “PLUG” when the robot is on the charger, and turns red below 6.20 V. Read the post.
- First battery life measurement. Unplugged on Sunday at 11:16 p.m., asleep: it switched off around 2:18 a.m., about three hours later. The following week measured it awake. Read the post.
What I learned
A silent robot is not necessarily a broken robot. And a voltage read by the motors while charging is the charger’s, not the battery’s. I will come back to this in a post about battery life.
The following week
- Measure battery life while awake, several times, so the gauge tells the truth.
- A real sleep, motors off, woken by an antenna.
- A name.