Biped: T0.8 "Mao"

This will be my mech like biped named "Troublesome", current version 0.8. Code name "Mao". Why Mao? Because it's 100% Chinese of course and since I live in the european union of dictatorship, the name of a dictator seems appropriate.

Mao's bracket arrangement is of my own design. (It used to be a 17 DOF humanoid.) It has 6 DOF per leg and when finished another 2 (or) 3 DOF in the head. It all depends on the total weight.

The "brains" of the biped will be an Arduino Mega and the power will be delivered by one power source (3s LiPo). Two 15A peak Buck converters running at 6v will power the servos, a 5v voltage regulator will take care of the sensors (ultrasonic, switch, infrared, accelerometer and compass) and the Arduino will be powered directly from the battery. I'm not sure yet if that will result in a lot of noise, but I have no weight to spare for an additional battery.


The programming part will be interesting. I will try to avoid inverse kinematics, because it's complex and I don't need accuracy. I will be using a set of fixed "rules" together with trigonometry to get the desired walking gaits and movements. In the prototyping phase the software will be running on my laptop connected to the Arduino via USB, after that I'll translate the program to Arduino code and then Mao will have to explore the world on it's own.



Mao may look quite intimidating as a first robot and it is. I'm not entirely sure myself why I thought this would be a good idea as a first robot. On the other hand, it's not exactly rocket science so what can go wrong? ;)

Will explore the world autonomously, eventually.

This is a companion discussion topic for the original entry at

Looking very good


I am working on a very similar robot.

Do you mind a little criticism?

I think the servos you’re using do not have enough torque. Specially in the ankle positions.

Won’t the U piece in the hip bend? you can use some spacers to make it a it more solid.

I am also using a laptop and arduino. Laptop code has the algorithms and arduino only sets the servo positions. I am using a bluetooth/serial adapter in the arduino to avoid having a usb cable. This way I have my bots roming around and I keep seated editing the code.


No I don’t mind criticism

The U piece does indeed bend, that will be fixed either with spacers or maybe aluminium panels.

As for the servos: as of yet I have no clue if they’re powerful enough. I hope so, the only alternative I could find were 15kg-cm servos which are a bit slower than the 996Rs and a bit more expensive. I can’t afford brand name servos, so if the servos are way to weak then my biped will become a quadruped. :wink:

** If you need more torque**


If you need more torque take a look at HDKJ servos. They’ve both wingless and winged servos. Usually cheap. Maybe a bit slower but I’m not sure.

There’s also a Spring RC that can handle 17Kg and a Turningy. The later one is what I’m using for my bot ankles.

hey, wait a minute here

isn’t this the same guy who just told me a walking biped robot can’t be built without a team of research scientists and a multi million dollar budget? Have you tried gyrscopes for automatic stabilization to keep this robot upright? I’ve been getting the gyro sensors for $20 each from Black and Decker cordless drills. Once the robot can stand up on it’s own and keep it’s balance, walking is really nothing more than putting one foot in front of the other. something like how a wind up toy walking robot moves it’s feet. Good luck with your project, looking forward to seeing more updates soon.

Stop thinking it can’t be done, and do it!

Size does matter!

This is a small biped, not a full scale robot. When size increases, everything increases. I might get away with my cheap 10kg servos, (It’s possible that they’ll stall or burn out, in that case I will make a quadruped. Or I’ll have to use servos that are at least 4 times as expensive and I need 12 of them in the legs.) if I should double the size of my robot  I at least double the weight of my robot so I have to have at least 3 or 4 times as powerful motors, which will use a whole lot more electricity meaning that I’ll need higher capacity Lipos, which will increase the weight even more, etc. Powerful servos, 52kg Herkulex for example, cost $$207.65. I now use $5 servos. $5 x 12 = $60, $207.65 x 12 = $2491,80! And then were still talking about quite a small robot that’s about 80cm tall.

If you assume that walking is nothing more than putting one foot in front of the other then you couldn’t be more wrong, walking on two legs is a difficult challenge. It takes a child more than a year to learn to walk, thinking that it is easier for a robot is quite the opposite.

Challenge accepted!

I don’t expect that it will take more than a year for me to teach a robot to walk.

We did some pretty cool halloween decorations using salvaged windshield wiper motors.  They were $10 each, and pretty powerful.  The weight is extremely important, and that’s why I recommend cardstock lightly coated in vinyl, or lightweight PET plastics (think the plastic bottled water comes in).

Meccano toy company makes a Meccanoid robot for around $400 that’s 4’ tall, and made of metal and plastics.  It’s locomotion style isn’t close to what I’m trying though.  I’m using a modified variation of Runbot for legs.  The weight factor is why I don’t believe InMoov will ever be able to walk, inspite of how impressive that project is otherwise.

My general understanding of the electrical system is that we don’t want to use more than 500mA at up to 36V to power motors, unless we redesign the circuit boards to handle more power, and that is using the H bridge method. Otherwise Arduino boards burn out above 5V at 500mA, so that’s the range of power that I am dealing with.  5-36 volts at 500mA for each point of articulation.  I’m trying to keep both weight and noise under control, so most of my points of articulation will be done with $1 motors, walking and the balancing trick though will be done with larger motors and the gyro sensors.

Bipedal walking is a challenge, but I’m not reproducing the Honda laboratory’s work with Asimo.  This is a much more scaled back and conservative locomotion style, more similar to the walking style of a wind up walking toy. You are right in thinking this is done with a combination of trigonometry and physics. The mass of the robot and the constant of gravitational force determines the rate at which the robot falls forward, and the foot needs to move at an upward angle to be in the location to catch it on the way down. This will involve some experimentation, and I will be surprised if I get everything right on the very first attempt. However, I don’t believe my plans reinvent the wheel, but rather are a conglomerate of several open source projects into one.  None of which are particularly easy, but I enjoy the challenge.

I do have the benefit of a degree in computer programming, prior experience of building a robot that functioned in either autonomous mode or by remote control (in only 6 weeks!), and professionally I operate a 5-axis industrial laser to cut prototypes and limited test run parts for manufacturing.

Original prototypes can definitely be made from cardboard, though most are coated in resin, epoxy, and paint when finished.  I plan to finish my droid in a light coating of vinyl.

Any news or updates on this

Any news or updates on this bot. I would really like to see a video if possible. I use the 996R servos all the time and they are pretty decent and should work for you. Let us know.

Not much

I’ve done not much with this robot, too many projects and other things to do. I did a few tests with my servos running at 6v and they get hot quite quick even when I run the servos (continuously) without load. Maybe I can lower the voltage to 4.5v without stalling the servos.

As for the software: I’m still convinced that simple trigonometry will work. I could program a simple simulator, but I prefer to look around and learn from the real world i.e. how humans walk and the relations between the joints when walking. I even got myself a Mindstorm 2.0 set to be able to quickly put together linkages and joints to confirm (or not) calculations. (And I like Lego, I just needed a reason to buy a set. :wink: )