Davy Jones Animatronic Tentacle Beard

Davy Jones Animatronic Tentacle Project from Pirates of the Caribbean

Project Overview

2020

After building my first pirate animatronic a couple years ago with my brother, which I have named "The Helmsman", I wanted to take a stab at a much more complex build for a very dynamic pirate character, Davy Jones. I am starting on this goal of recreating some of the famous scenes from the Pirates of the Caribbean movies that feature Davy Jones by beginning to prototype his tentacle beard.

I will break down this tentacle project by the following categories to help explain how I went about working through each challenge:

Research- I started by doing research to gain insight into the overall character of Davy Jones and how his tentacles behave and reflect his personality.

Design & Prototype- Based off of the findings from my research, I was able to design, manufacture, and assemble the tentacle mechanism as a three-stage, cable-controlled, servo-driven mechanism where each stage has the ability to move in 2 DOFs independently from one another. I also designed a motor mount rig to house all of the motors in a compact packaging space so as to fit inside any future character head designs.

Electrical/Controls- The mechanism consists of 12 servos being controlled by an SSC32U servo controller board. The servos are split across two rails on the controller and are powered with two separate power supplies.

Animation- I used Visual Show Automation (VSA) to keyframe animate the different stages for both tentacles in sync with various movie clips from the Pirates movies.

In the future, I hope this prototype will lead to many more projects that will eventually result in a life-sized Davy Jones animatronic figure.

Research

Research, Research, and More Research

Goal

My research goal was to obtain an accurate, realistic, and life-like idea of the overall look and feel of Davy Jones's tentacle beard.

In order to fully bring a story as tragic and epic as Davy Jones's to life, it is most important to learn to think like the character. In my research, I wanted to try to understand how Davy Jones feels, what he wants, what he needs, and what psychological mechanisms make him the character he is.

This way, hopefully, the animatronic tentacle beard and thus the Davy Jones animated figure can really come to life and reflect the nature of the character as he was brilliantly created and portrayed in the movies. The research is meant to inform the design choices and refine the approach so the story really tells itself when all is said and done.

I learned from my previous build of my first pirate animatronic, "The Helmsman", that research helps me shape my expectations and vision for a project. I tend to start out with a plan or rough idea of how I think I can best bring a certain character or scene to life, but when I begin researching, reading forums, and looking at others who have done similar projects, I find my expectations change rapidly and my approach evolves along with that too.

Approach

Before getting started on any attempt to design/prototype this animatronic tentacle beard, I wanted to answer some basic questions:

1. What are the basic dimensions of Davy Jones's tentacles? This overarching question can be broken down into the following smaller questions:

a) How many tentacles does Jones have?

b) What are the overall lengths of the different tentacles on Jones's beard?

c) What are the overall widths of the different tentacles in their non-tapered section on Jones's beard?

d) What is the length of the tapered section of all of the different tentacles?

e) What will be the width of the actual mechanism itself ( i.e. the dimensions of the actual mechanism that will be housed in the silicone tentacle skin)?

f) What is the slope of the taper for the different tentacles?

2. What will be the basic working model for my tentacles? Again, this overarching question can be broken down into several smaller questions:

a) How many independently controlled stages will there be?

b) How many degrees of freedom will each stage have?

c) What are the motor requirements for the design?

3. When animating the tentacles, which forces are important to simulate in the animation to really bring the squid-like characteristics of this tentacle beard to life? How do I preserve Davy Jones's cruel, tormented personality in the behavior of the tentacles?

Results & Findings

I compiled all of my detailed notes and findings from my research in a document that explains how I arrived at the best answers I could for some of the above questions. I have linked it just below. I will also summarize my answers neatly below in the same order as the questions are listed above and add some details where necessary:

My preliminary research for all basic tentacle parameters (includes tentacle geometry analysis and key animation styles/tentacle behavior): click here

PDF detailing some of my rough hand sketches for different tentacle designs in terms of overall look and shape: click here

PDF showing my approach to doing hand calculations to determine different mechanical requirements for the tentacle design: click here (NOTE: I am neglecting the inertia of the tentacle in these calculations. Loads will be higher in reality and motor requirements will therefore need to be chosen accordingly with this assumption in mind).

1. What are the basic dimensions of Davy Jones's tentacles? Let's answer this question by answering several smaller questions:

a) How many tentacles does Jones have?

Davy Jones has 46 tentacles in all (per Industrial Light and Magic).

b) What are the overall lengths of the different tentacles on Jones's beard?

If I am assuming the two chin tentacles and the two lip tentacles are the same length, then it appears they are approximately 24" in length. I could see anywhere from 24"-26" being a reasonable length to make the lip/chin tentacles. A good compromise might be 26" OL for the two lip tentacles and then 24" for the two chin tentacles.

c) What are the overall widths of the different tentacles in their non-tapered section on Jones's beard?

So far in my research I have primarily analyzed Davy Jones's chin and lip tentacles. They appear to have an overall width of approximately 30mm in the non-tapered section and a 35mm depth on the side where the tentacle suckers are.

d) What is the length of the tapered section of the different tentacles?

Based on my estimate that 2/3 of the tentacle OL is non-tapered, this means approximately 1/3 of the tentacle's OL is tapered. For a 24" OL tentacle, the tapered section would be around 8" long.

e) What will be the width of the actual mechanism itself (i.e. the dimensions of the actual mechanism that will be housed in the silicone tentacle skin)?

I am using 21mm for the tentacle mechanism itself in the non-tapered section. If I were to use 1/4" thick silicone for the tentacle skin, I would end up with an overall width of around 33mm for the tentacle which I am fine with. Of course the depth would be more than 33mm because of the suckers on one side of the tentacle.

f) What is the slope of the taper for all the different tentacles?

I am using a taper slope of -1.625mm/inch when starting at 21mm and ending at 8mm over the 8" tapered length. In order to figure out this slope I only needed the starting point width (21mm), the ending point width (8mm), and the tapered length (8"). The starting point width of 21mm is the width I am using for the non-tapered tentacle mechanism width/diameter mentioned above. The ending point width of 8mm is the smallest I believe I can feasibly make the bottom-most vertebrae of the taper and still maintain accuracy in my 3D printing and such.

2. What will be the basic working model for my tentacles? Let's again answer this question by answering several smaller questions:

a) How many independently controlled stages will there be?

After studying the pirates movies and looking at various ILM (Industrial Light and Magic) models, I think I can still achieve the proper motions in the animation while preserving the true character of Davy Jones with three independently controlled stages. The first 8" or so will be stage one, the middle 8" will be stage two, and then the bottom 8" where the tip is will be stage three.

b) How many degrees of freedom will each stage have?

Each stage will have two DOFs, and each DOF will be orthogonal to one another. To clarify a little bit, the lip tentacles will have all basic degrees of freedom except the ability to twist along their center/core axis. I do not want the tentacles to be able to twist along their core, but I do want them to be able to move about the other 2 basic axes/planes of motion in 3D space.

c) What are the motor requirements for the design?

I did some detailed hand calculations (see PDF above) to try and get a good estimate for how much motor torque/speed would be needed in order to animate these tentacles in a way that accurately reflects the behavior of Jones's tentacle beard. It should be noted I have neglected the inertia of the tentacles in these calculations. Therefore, the loads in the system will be higher at times than my calculations reflect. Ultimately, I will need to size the motors to be more powerful than what my calculations will reflect. All things considered, I decided on the Hitec HS- 7950TH servos for the first stage of the lip tentacle, Hitec HS-5585MH servos for the second stage, and Hitec HS-5645MG servos for the third stage.

ILM published a quick one page document where they detailed their approach to performing a rigid body dynamics analysis on Davy Jones's tentacle beard when they were animating the character for the Pirates movies. The article mentions several aspects of the animation for Davy's tentacle beard that make the beard really come to life.

It was important to them for the tentacle beard to "[behave] like that of an octopus, but still [present] the dynamic motion of the character’s performance". When animating these tentacles I wanted to make sure to bring the octopus-like quality to life without suppressing the heart/personality of Davy Jones that needs to be reflected in how the tentacle beard behaves.

The article also mentions how collision of the tentacles with each other and other objects was important. John Knoll coined a term for a certain type of collision reaction force known as "stiction". This simply refers to the "wet/limp noodle" effect where the tentacles collide but sort of grab onto each other and spring off of one another.

Gravity also plays an important role in animating the tentacle beard too. When Davy's head moves, sometimes the tentacles lay on his chest as they are pulled down by gravity. Buoyancy could also play an interesting role here.

Another great video by ILM showcases a sort of "tentacle calisthenic" video of different standard tentacle poses that could be programmed and used to more easily animate the tentacle.

Design & Prototype

Goal/Getting Started

My overall design goal is to take a previously designed two-stage tentacle mechanism, scale it down/change the design where necessary to fit my application, design a third stage for the mechanism, motorize the control rather than use hand controllers, and design a motor mount test rig.

Regarding the general multi-stage tentacle mechanism model, I have not created or invented this mechanism. Instead, I am expanding upon an original concept from Joshua Vasquez. Joshua has done a fantastic job of laying out how his mechanism works and designing his own hand controllers.

The basic design principles behind multi-stage tentacle mechanisms are explained very well in the linked resources below. This information really helped me figure out how I wanted to tackle my goal of designing a three-stage Davy Jones tentacle.

Joshua's two-stage tentacle mechanism tutorial on Hackaday

Joshua's youtube videos of his two-stage tentacle

Joshua's Flickr build log for the two-stage tentacle

Brian Poor's tentacle mechanism tutorials

Designing the Tentacle Mechanism

In order for my Davy Jones tentacle to achieve fluid motion and look as life-like as possible, it needed to be fixed rigidly at its base. I started by designing a baseplate that was intended to act as a rigid connection between the base of the tentacle mechanism and the motor mount test rig.

I made sure the baseplate could route all control cables and any cable conduit for each stage of the tentacle mechanism. I added slots for the second and third stage cable conduits to be able to route through the baseplate and into the tentacle. I was mainly concerned with making sure the slots provided adequate room for the conduit to shift as the tentacle moved into different poses.

Since the baseplate is intended to function as a rigid connection between the tentacle and the test rig, I decided to laser cut the part out of 1/4" delrin. Delrin is immensely durable and tough.

Once I finalized the baseplate design, I moved on to optimizing the first and second stage vertebrae designs to fit my application. I scaled down Joshua's original tentacle vertebrae design to the correct dimensions for my application based on my research of Davy Jones's tentacle geometry. I had to pay careful attention to make sure the slots in these vertebrae could house two cable conduits each with adequate room for the conduits to shift and bend as the tentacle moved. I also adjusted the radius of the hole pattern for the control cables. I wanted to optimize the design for servo control rather than hand control. I decided to have the control cable hole pattern sit on a slightly larger circle (relatively speaking) in order to reduce the required motor torque.

It is important to note when adjusting the radius of the control cable hole pattern, there is a trade-off between torque and range of motion. This means I slightly decreased the range of motion of the tentacle by increasing the hole pattern radius, but the difference did not affect my animation requirements.

I decided to laser cut these vertebrae out of delrin as well since they would be encountering some internal bending moments during different tentacle poses. However, I decided to make these parts using 1/8" delrin to try and get as much resolution as possible in the tentacle mechanism's movement.

Then, I moved on to figuring out a way for these vertebrae to fix rigidly to the tentacle core without moving or spinning around. It is important for the tentacle's vertebrae to remain fixed relative to the tentacle core to ensure the reliability and accuracy of control. I wanted to design a separate part using aluminum so I could have threaded holes machined into it. This way, set screws would be able to provide the necessary friction force for securing the vertebrae without cracking or damaging it. I designed the part to be bolted to the tentacle vertebrae with two bolts. I also made sure the part would not interfere with the control cables routing through the tentacle.

Lastly, I designed from scratch my own "tapered tip" for this tentacle's third stage. I decided an 8" long taper starting at the mechanism's full width and ending at 8mm would best represent the character of the tentacles in the motion picture.

I designed four uniquely tapered vertebrae that each took on the shape of a truncated cone. I added a cylindrical cutout in each vertebrae for the tentacle core to pass through. I also added lofted cutouts for the control cables to pass through. My last task was to add threaded holes for set screws.

I wanted to experiment and see if a functional vertebrae could be 3D printed with ABS. Since I was using ABS, I decided to place four through-holes equally spaced around the outside of the vertebrae. This way heat set inserts could be set into the ABS for set screws to hold the vertebrae firmly in place.

I initially tried to secure each tapered vertebrae with two set screws but found out the design wasn't sufficient to prevent the vertebrae from spinning/moving around. Thus, four equally spaced through-holes in the final design.

After finalizing my designs, the next step was to move to acquiring all off-the-shelf components necessary for building the tentacle as well as 3D printing/laser cutting/machining all of my custom designed parts.

When ordering my off-the-shelf components from various places online, I found it to be very cost-effective to place a single, large order rather than multiple, small orders. This ultimately saved money on shipping. Thus, the importance of doing as much planning, researching, and designing as possible before trying to order parts to start building. Of course, when prototyping a design and doing some of your own R&D, there will always be some amount of cost associated with R&D which is unavoidable.

Baseplate design (above)

Tentacle vertebrae design

Third stage/tentacle "tip" design

Build Photos

Motor Control & Test Rig Design

Most of the snake arm/continuum robots/tentacle mechanisms I see online utilize hand controllers to motorize them. These hand controllers can produce extremely life-like and fluid movements for tentacle mechanisms. However, I wanted to see how much realism I could achieve with a motorized design.

With that in mind, I wanted to build a rig to house the motors for both lip tentacles that would resemble a potentially feasible figure head design in the future. I decided to dedicate three different planes stacked on top of each other for the twelve motors that are being used between the two tentacles. The first stage motors for both tentacles are mounted on the bottom plane of the rig, the second stage motors for both tentacles are mounted on the middle plane of the rig, and the third stage motors for both tentacles are mounted on the top plane of the rig. Three inches separate each plane in this rig, but in reality this could be reduced to 2".

Each plane is stacked on top of one another and mounted to the rig with bolts. This means if I needed to access the bottom or middle plane for whatever reason, I could take off the appropriate bolts and lift the successive plates off the rig in order to access the desired plane.

I made sure to include a mounting panel behind the rig to mount the SSC32U servo controller so all the servos could attach to the controller. This way a cable can run from the controller to my laptop to animate the tentacles.

Lastly, I designed some brackets to be 3D printed which are bolted to the rig. These brackets act as the termination point for the extension springs that are being utilized as cable conduit in this design. I added a cutout to the part for a bicycle brake cable ferrule to be press-fitted in so as to transfer stresses off the 3D printed part and onto the off-the-shelf aluminum ferrule. The control cable passes through the bracket and terminates on the motor pulley.

I hope to continue refining the motor mount/test rig design to be used as the functional character head design down the road.

Motor mount test rig design

Electrical/Controls

I used a 7.5V 50A switching power supply to power the first and second stage servos for both tentacles on the test rig since these servos are high voltage, high torque motors. The third stage servos for both tentacles run on 6V, so those servos are being powered by a separate 6V 5A power supply.

All of these servos are connected to a single SSC32U servo controller. The eight servos that are running on 7.5V are hooked up to the VS1 rail of the SSC32U while the four servos running on 6V are hooked up to the VS2 rail of the SSC32U. In order to use two separate power supplies, I had to remove the VS1=VS2 jumper on the SSC32U. Additionally, the SSC32U is designed such that when VS1 voltage is above 5.3V (not counting temporary drops) the logic of the board can be sufficiently powered. So, my 7.5V power supply connected to the VS1 terminal also powers the logic of the servo controller.

The 7.5V power supply is connected to the SSC32U via a positive (red) wire and negative (black) wire connected to the positive and negative VS1 terminal slots, respectively. This wire is 18AWG stranded wire, which is the largest stranded wire the SSC32U can accept. I have not tried to fit solid wire into these terminals. It might be possible for the terminals of the SSC32U to handle 16AWG solid wire, but I have not tested that theory. The 18AWG wire is rated to handle 7.5A continuously, which is suitable for my current application.

The 6V power supply is connected to the SSC32U via a wiring harness which accepts a 2.1mm x 5.5mm barrel jack. The positive (red) wire and negative (black) wire of the wiring harness are connected to the positive and negative terminals of the VS2 rail, respectively. The wires on the wiring harness are also 18AWG wire. These wires appear to be solid wire as opposed to stranded wire.

It should be noted that each rail of the SSC32U is rated for no more than 5A of continuous output and 15A-16A of peak current at any given time. Additionally, the standard pins on each power rail used for SIGNAL/VCC/GND of the RC servomotors can support at most 3A continuous themselves (of course, the board itself cannot provide 3A x 16 servos on one rail). I needed to make sure to draw no more than 5A continuously and/or 16A at peak from each power rail during the animation of the tentacles on the rig.

It should also be noted that if more current is needed than the cabling/power systems can handle, then voltage will drop and the wires themselves could begin to heat up. If voltage drops low enough, the coils in the motors won’t have enough energy to actually turn/hold a position. This can cause temporary, uncontrolled, and erratic behavior from the motors. In any case, as long as the current demands from the load stay within the SSC32U constraints, then the 18AWG wire will not be at risk as the wire can handle higher loads than the servo controller can.

In order to avoid placing the SSC32U and potentially any cabling under too heavy of a load, a couple approaches can be taken:

Servo movements can be staggered by a few milliseconds. The spikes on initiating moves under loads aren’t all that long in duration. So instead of having 30 servos move at once, you have 30 servos start their moves over 5-10 ms. This is barely noticeable to most humans but also greatly reduces peak currents.

The amount of servos per rail can be reduced and then spread out over more SSC32Us. If more SSC32Us are used, it should be noted that a thicker wire will be needed to support the loads coming from the power supply. This thicker wire should then be spliced with the 18AWG wires which will connect to the various SSC32U being used. The following image depicts this type of setup.

For more helpful information on the power circuit for this project, I highly recommend going to read my forum post where the leader, Scharette, does a marvelous job detailing out potential issues and solutions for this project. Most of my information in this step has come from this forum post.

Animation

When I built my first pirate robot, The Helmsman, I used Visual Show Automation (VSA) from Brookshire Software. This software has enormous capabilities for allowing simultaneous control of dozens of servos, DMX equipment, audio tracks, and even video files during a show/performance. This software is by far one of the most convenient and easy-to-use control softwares for a robot project where an animation profile is desired in which servos or DC motors need to move in synchronization with time/audio/video cues or in a live capture situation.

The Skinless tutorial is an excellent step-by-step tutorial on how to use VSA to control servos and how to import an audio file to synchronize servo movements with said audio file.

One important lesson I learned while creating the animation profiles for the tentacles was to make sure the bandwidth out of my COM port was sufficient given the demands of the profiles. Depending on which types of devices VSA is controlling, a noticeable slow down or glitch can be observed if too many events are occurring at one time. The VSA help manual says that if too many events are occurring at one time out of one COM port, there are three approaches that can be taken to increase bandwidth in order to get rid of the slow down:

1. Some devices can be taken off of a COM port and then moved to another COM port.

2. The frame rate can be increased in the COM port settings (F3) in VSA.

3. The Baud rate on the controller and the COM port in VSA can be increased (F3).

For part of my animation, I was trying to send nine SSC32 servo commands at once, and I noticed a slow down during the animation. The VSA timeline appeared to lag. I decided to increase the baud rate of my SSC32U controller and COM port from 9600 to 38400. This baud rate was sufficient to provide enough bandwidth given the demands of my animation profile.

The animation profiles I created contain a calisthenics portion, which includes different standard tentacle poses, and several animated scenes from The Pirates of the Caribbean film series shown below.

Movie scene animation (organ playing)

Movie scene animation (Liar's Dice/Key Scene)

Tentacle mechanism in action side-by-side with the movies

In the Future

Stay tuned...

Moving forward, I plan on prototyping different mechanisms for the rest of the tentacles I want to include in the Davy Jones tentacle beard animatronic. Eventually I will move on to trying to design a character head, different mechanisms to go inside the head, different mechanisms for the full body, and also some shell designs for the figure. Stay tuned as I continue to develop this project and strive for a lifelong dream of mine to design and build a full-sized animatronic Davy Jones figure!

Jason Collins| Mechanical Design & Analysis