2025-04-09, 07:23 AM
(This post was last modified: 2025-04-09, 07:37 AM by seandepagnier.)
I used pypilot quite a bit on my kracken 33 trimaran. It performed well at all boat speeds for me, but there is always room to improve.
The biggest improvement is I used a high speed actuator that moves 150mm/s. This rotates the rudder 30 degrees per second. Most corrections are for fractions of second but the fast movement makes a big improvement. I reduced the servo.period to 0.25 and tweaked some other settings to make it react quickly.
Slower actuators that moved half this speed were fine in lighter wind but never steered as well once I exceeded 8 knots. This might be the issue with other autopilots too though pypilot likely outperforms raymarine by a good margin anyway.
There are not many direct comparisons to other pilots but so far in all but 1 or 2 cases the results are that pypilot performs better. There are a lot of variables involved. I expect the more advanced pilots you listed to potentially have an edge in some conditions but without their source code or algorithms it is difficult to say. I have received reports that pypilot outperformed a b&g pilot. Again it really depends.
As for racing in big swells... This is tough. Because relying on the autopilot could even be dangerous. My trimaran was built in 1973 before they put enough buoyancy in the outer hulls so I was nervous to push it, especially downwind as this is a known capsize risk when it is surfing and a hull digs in. I felt comfortable sailing 12 knots on a reach and occasionally would go 14-15 knots but my sails were original from the 70s and had poor shape (making autopilots for years kept me too poor to upgrade) so the performance was limited.
downwind I would typically slow down to 9-10 knots with swell. Newer designs of trimarans should be much safer. It would be interesting to perhaps have additional safety logic in place but I am not really sure what is best. Perhaps a rocket on top of the mast can be used to right the boat if needed. But more seriously an additional computer with sensors that can do something to prevent capsize. This was always my biggest fear as I managed to bury (submerge a hull completely) or roll to 45 degrees in the gulf stream.
The pypilot repository is for the autopilot and has more recent improvements but recently not many unfortunately. As for tinypilot this is just the base operating system built on tinycore linux so it has not needed many updates.
The biggest improvement is I used a high speed actuator that moves 150mm/s. This rotates the rudder 30 degrees per second. Most corrections are for fractions of second but the fast movement makes a big improvement. I reduced the servo.period to 0.25 and tweaked some other settings to make it react quickly.
Slower actuators that moved half this speed were fine in lighter wind but never steered as well once I exceeded 8 knots. This might be the issue with other autopilots too though pypilot likely outperforms raymarine by a good margin anyway.
There are not many direct comparisons to other pilots but so far in all but 1 or 2 cases the results are that pypilot performs better. There are a lot of variables involved. I expect the more advanced pilots you listed to potentially have an edge in some conditions but without their source code or algorithms it is difficult to say. I have received reports that pypilot outperformed a b&g pilot. Again it really depends.
As for racing in big swells... This is tough. Because relying on the autopilot could even be dangerous. My trimaran was built in 1973 before they put enough buoyancy in the outer hulls so I was nervous to push it, especially downwind as this is a known capsize risk when it is surfing and a hull digs in. I felt comfortable sailing 12 knots on a reach and occasionally would go 14-15 knots but my sails were original from the 70s and had poor shape (making autopilots for years kept me too poor to upgrade) so the performance was limited.
downwind I would typically slow down to 9-10 knots with swell. Newer designs of trimarans should be much safer. It would be interesting to perhaps have additional safety logic in place but I am not really sure what is best. Perhaps a rocket on top of the mast can be used to right the boat if needed. But more seriously an additional computer with sensors that can do something to prevent capsize. This was always my biggest fear as I managed to bury (submerge a hull completely) or roll to 45 degrees in the gulf stream.
The pypilot repository is for the autopilot and has more recent improvements but recently not many unfortunately. As for tinypilot this is just the base operating system built on tinycore linux so it has not needed many updates.