The RobotShop Learning Center is intended to provide both new and seasoned robot builders with the tools, information and contacts they need to buy, build or troubleshoot ever increasingly complex robots. The Learning Center itself is a dynamic tool which will grow and change over time. Feel free to discuss Learning Center content as well as content you would like to see added in the future.
If you are interested in making a contribution to a section of the Learning Center (an article, project ideas, a dynamic tool, news items etc.) please contact Technical Support via the Support Center Subject: Learning Center.
Welcome to the RobotShop Forum. Your post is indeed in the right place for the question and your feedback is very much appreciated. You are correct that in both of the equations the last term should be positive. The equations will be corrected shortly on the website. When making the Robot Leg Torque Tutorial, we quickly realized that it could only take into consideration one configuration and one type of walking gait. We chose to go with the most popular (hexapod, with three 3DOF legs always in contact for static stability). If someone is working with a dynamic gait, or a quadruped (or an octapod?) that they can take a similar approach. What type of robot are you working on?
To see what the torque at each joint would be under worst case scenario, you should only take into account each individual joint (considering the rest as rigid links) - essentially you are left with two “sticks” connected to each other by a pivot, supported at the ends. You would need to re-do the calculation for each joint separately (one side supported by X legs and the other supported by X/2 or other fraction). This assumes that the body is symmetric (would yours have 5 legs on either side?). Your first issue will be what type of actuators to select. The approach for designing a 10 legged, 200kg robot is definitely unique - you may want to take a slightly different approach and select an actuator technology and then determine the load they can support. I can only image high power linear actuators, pneumatics or hydraulics would work. Would you also look to make each leg 3DOF? A 2DOF leg may prove far less expensive and give similar results. Do you have a sketch you can scan or something to give an idea as to what it would look like?
If you try to calculate the torque about the knee, the two sides cancel to give you zero - intuitively this cannot be true since two “sticks” are not self-suporting. Instead of adding the torque between the two sides, consider the torque needed to support either side (i.e. consider the other side to be fixed). The torque in your scenario would give LW/2 for either side.
There is not much information online about optimizing legged robots - most of what can be found is written by PhD students working on one aspect of the robot. Feel free to bounce ideas off the community here. Why choose 10 legs?
There are not many walking robots out there yet, so even a functional custom hexapod is still interesting/appealing to the community. It’s obviously critical to design, test and refine one leg before producing 9 others that are identical. Looking forward to seeing your project progress (pictures always spur more conversation). Gotta have flashy headings too
Thank you for activating my membership.
I found your excellent site whilst attempting to check some leg torque calculations of my own.
I have a number of queries about the equations, but the most important one is as follows:-
In the equation for “torque balance around the LEFT FOOT”, the last term “+2N2 …” is a positive term which I assume is because the force N2 is the ground reaction acting upwards, resulting in a positive (counterclockwise) torque.
However in the KNEE and HIP torque equations it is negative in both cases ( “-2N2…” and I don’t understand why it has changed sign.
Please could you explain?
I hope this is the best place for this question.
If not, please could you tell me where to post any further queries about this tutorial.
Can I now please pose a question, that results from me looking very simplistically at the “corrected equation” for the torque around the knee (although my result for the hip equation is similar)?
I set a simple scenario to test my understanding and the result has confused me. I am sure I am missing something obvious and I would appreciate your help:-
I decided to set all the lengths to be L ( L1 =L2 =L3 = L)
I then set all the angles to be equal at 0 degrees - so I have effectively got a flat robot !
I have then simplified things further by making W1 =W2 = W3 = 0
I set the weight W4 to 100 Kg so therefore N1 is 50 Kg and N2 = 25 Kg
When I enter these values into the knee equation it appears that T1 is zero.
I then re-checked making all the angles EQUAL, but NON-ZERO, and realised that the actual angle is irrelevant - the result is still T1 = zero.
I would welcome your assistance.
I think your choice of a hexapod as an example in the tutorial is highly sensible and very useful.
I am at present in the early design stages of a long term project to build a large decapod robot. I have a target mass for the entire robot of 200Kg and I am examining different leg/joint configurations. I have various design constraints not least of which is cost, and as a result I have decided to make all mechanical components from scrap. This includes the actuators and it was whilst trying to optimise an actuator design that I started this torque check.
I look forward to your response.
Regards
I think you have understood exactly my scenario - my simplified check model is essentially “two sticks” connected by a pivot
In my scenario 1 stick is length L and the other stick is 5L
The knee is at the pivot (i.e. 1L from the left hand side)
and the weight W4 is in the middle ( i.e. 3L from the Left Hand Side)
If I simply assume that there are 2 legs, 1 on each side, I am assuming that they will each see the upwards force = half W4
But when I then substitute these values into the knee torque equation I still seem to end up with T1 = zero - which can’t be correct !!
So I am still very confused - I think I am missing something very obvious. I just can’t see what to do next.
I would appreciate further comment if you have time.
Thankyou
Concerning my decapod, I am considering designs with 2 or more leg sections with at least 1 section being telescopic. I am considering up to 5 DOF for the leg (when rotation at the hip and foot is considered). I am indeed looking at linear actuators as my first choice, with electric power. I have already successfully tried making my own ACME screws and nuts and have successfully driven them with my own brass spur gears. The ten legs will all be independent and I am at present tending towards a symmetrical design. What I have not yet said is that it is a legged/wheeled design with a steerable wheel at each foot which can either be powered for wheeled motion or locked for walking. I recognise that with electric linear actuators the walking speed will be very slow, but for my project I am not worried about a fast walking speed.
I am present developing a full 3D model (programmed in C and Open GL) and am merging this with a “dummy” terrain" so that I can investigate leg configuration for walking.
I now understand - I knew it was intuitively wrong but I couldn’t see my way through - I was trying to make things too complicated - Many thanks.
One reason I opted for 10 legs was to do something different as a long term challenge- I didn’t want to simply do something very similar to anything already designed. I have read hundreds of research papers and PHd theses and decided 10 legs was unusual. Of course I realise that it may also be impractical, due to overall weight etc…
A second reason was that, having first thought about a hexapod or an octopod, I decided to go 2 legs further, both for mission redundancy and also with a view to possibly using several legs as manipulators in the future. I am starting with 10 identical legs but I am also considering the possibility of 4 of the “legs” being multi-segmented snake type limbs. If these were wire actuated then the actuators could be placed more centrally (ie in the body) and hence slightly reduce torque requirements for the other actuators. I have so far modeled a five segment snake with 10 actuators and this is now included in my overall 3D model.
I have now read all through the forum and I certainly intend to make use of (and hopefully contribute) to the forum.
Are further typo editing comments re the leg tutorial of use for any . I am happy to email them if that is preferable.
