Friday, January 16, 2015

Autopilot Guidance

To say the boat owner was upset was an understatement. The boat had nearly thrown himself and it's occupants into the drink. Without seeing the event I already had a good idea about what had happened, and I've been on a boat in the recent past that had done the same thing with me on board. It was scary to say the least to have the boat tipped 45 degrees on its side in a high speed turn.

I patiently listened to the story. The boat was under autopilot control and traveling around 40mph. A crab trap buoy was spotted ahead and the autopilot was disengaged. The owner steered around the  buoy and re-engaged the pilot and this is where things went awry. The boat was about 40 degrees off course, way off the course line and still traveling at about 40 kts. There are ways to mitigate this event, but not alway completely and sometimes with a small price to pay.

Here is part two of the discussion. In the news this week was the story of a former NFL player who fell overboard landing a fish in the dark while his boat was under autopilot control. The boat wasn't moving quickly but was too fast to catch. The individual was able to swim the 9 miles back to shore. He is very lucky boy. The boat was found later off Grand Bahama Island. 

This discussion is generic. Each autopilot system although different, have a lot of common design elements.

There are two things the autopilot does for you. One is to maintain a steady heading, this is the "auto" function. What I mean by steady heading is the autopilot's job is to keep the bow pointed in one direction. What the auto mode can't do is to compensate for the effects of wind and waves.

To illustrated this let's imagine a boat in auto mode traveling along and there is a steady 10 kt breeze blowing onto the port side. What will happen is although the heading stays the same the boat will be steadily pushed to the starboard as it moves along.

Good, now consider we are using the autopilot to track to a waypoint. The autopilot now has a new job. It has to stay on the course line and deal with other forces such as wind and waves. With our same 10 kts blowing onto the port side the computer has to calculate a crabbing angle to compensate. What you end up with is upwind travel direction changes to stay on the course line.

Let me do a quick summary. When tracking to a waypoint the course computer's job one is to keep you on the original course line, not to make up a new one. So what happened to the owner after steering around a crab trap buoy, and re-engaging to autopilot was he was about 40 degrees off the bearing to waypoint, and was now some distance away from the original course line.

The first thing the autopilot will do is to steer the boat back onto the original course line and this is what it did. I wasn't there but I'm sure the combination of the boat's speed and abrupt slewing of the boat as it turned caught him unaware. This can be a very dangerous combination of events.

I'm going to digress a bit. If the boat had been a 32 ft Grands Banks trawler lumbering along at 8 kts, going around the crab trap buoy the effect would have been the same but in very slow motion. The wine in the glass would have barely moved. 

I'm now regularly installing autopilots on boats that are easily capable of 70 kts or more, and most of these are step hull vessels. This is important because although there has been a vast improvement in the design and stability of these boats, at very high speeds any abrupt turn can have tragic consequences.

There are a couple of other things I'm going to throw into the mix. The first is response rate. This is how often you ask the autopilot to make changes. Once every few seconds? Twice a second? Or somewhere inbetween. The faster the response rate the better the course keeping. What you always want is the slowest rate that keeps a good course.

Having it too high uses a lot of power, and can be dangerous under some conditions. Consider this scenario. You're traveling along in 4' seas and the waves are coming from the starboard side at about 45 degrees. If you were hand steering what would happen is you would enter the wave, the bow will naturally fall off to port, and on the other side the bow would fall off to starboard.

Let's set the autopilot with a fast response rate and do the same thing. What happens is the bow doesn't fall off to port, but instead the boat tips to the port. On the other side of the wave the boat then tips to the starboard side. The minimum is you make the crew woosey with the motion, and the maximum is if the waves are large enough the boat could possibly capsized under some conditions.

Lastly the gain stuff. I define this loosely as when the autopilot makes a change (the response rate controls how often) how aggressively does it make a change? In older pilots this is more set, and in newer ones this is more adaptive. At 5 kts you have to add a lot of rudder to make something happen, at 70 kts it's just a tiny amount of rudder that is used. 

Now for Bill's rules to have a happy relationship with your autopilot. 

Rule one. Never leave the helm of your boat under any circumstances while the autopilot is engaged! If you need to leave the helm, disengage the autopilot, bring the boat to a complete halt, look very carefully around for hazards, and then go. The other rules 5 & 7 will bite you if you don't

Rule 2. Pay attention at the helm while the autopilot is in control. Just to be clear this doesn't mean drinking beers, listening to tunes and chatting with your buddy while you're screaming out to the fishing hole at high speeds. These are machines. They can break or malfunction. You should always be prepared to take immediate control of the boat. Keep in mind that 60 kts in only one second you have traveled over 100'. 

Rule 3. When engaging an autopilot to track to a waypoint be very close to the course line, pointed in the general direction, and traveling at moderate speed. Once you have acquired the course gradually throttle up to your desired cruising speed.

Rule 4. Periodically check the condition of the gear, and know where it is located. This is the course computer, compass, rudder actuators, and hoses.

Rule 5, Know thy compass and understand the impact a bucket of chain or a tool box next to it will have.

Rule 6. Keep the response rate at the lowest setting needed to keep a reasonably straight course.

Rule 7. Depending on the boat in anything bigger than 4' seas at the most you should be doing the steering. Although this will depend on conditions and wave period, don't be in the position of discovering belatedly the the autopilot can't deal with the adverse conditions.

Rule 8. Read the manual carefully, twice.

Now a bit of an editorial for the manufacturers of autopilots. The quality of autopilot documentation varies from excellent to good. What really isn't discussed well by anyone is the set up for autopilots that will operated at the now much more common higher speeds. There has been a lot of effort for the good in making them more adaptive and self learning, but too often this is for the majority of the market and not always for the extremes. The deeply burried in the menu set up functions are often not well documented if at all. I think this needs to be improved. Yeah I know you don't want the masses to play with these, but there is me, and I worry sometimes and feel I need to know more about how they really work.

The wake up call from your autopilot should not be the reef you just struck while you were asleep at the helm. I thought an interesting safety feature for an autopilot might be a lanyard and kill switch similar to the ones used for outboards. Wait a minute, I don't think I've ever seen a boater actually use one. I did find an interesting little sensor for about $50 that could tell if you were at the helm or not. Leave the helm and a relay can kill power to the autopilot. No, someone will find a way to beat the system. Darwin will just have to deal with culling the herd. 


  1. An excellent primer. In my own (sailboat's) fitting out, I have had reason not only to consider the proper role of an autopilot, but also its proper use. I dislike integration in the sense of setting the AP to steer for a waypoint; I prefer to steer to a compass course for a given period of time. If I end up short or long, or some distance from the waypoint I haven't input, I learn about set and drift of currents and (in my case) deficiencies in my sail trim.
    The lanyard kill switch is an interesting idea and simple to execute. Better, however, is the operator understanding that if they have to hit standby, that they had better line up manually with their waypoint at dead slow BEFORE re-engaging the AP, which, after all, doesn't care that your drink just got spilled. It's only doing its single-minded job! Oh, and in reading carefully, I note that the second Rule 6 should be Rule 8.

  2. Rhys, thanks for picking up the typo and the thoughts. My third editor eye wasn't working well that day I guess.


Note: Only a member of this blog may post a comment.