The following is a list of all the major work which was done on Spring Flamingo from the time it was decided to build the robot until the robot worked reliably and was demoed for a film crew. Not included is the actuator prototype which was done as an undergraduate thesis by Mike Wittig during 1995/1996 and was almost complete before this list starts.
In total 239 days where spent working on the robot. How these break down is listed below. Note that the break-down adds up to more than 239 (262). This is because some days were long and lots got done and some days a couple people worked on the robot.
As of late August, 1997, most of the work is now concentrated on walking algorithms. The robot is working fairly well with a mean time between failures of over 15 laps, and increasing.
Simulation, Walking Algorithm: 49 (19%)
Mechanical Design: 44 (17%)
Machining: 42 (16%)
DSP Development Software: 26 (10%)
Assembly: 24 (9%)
Testing, Prototyping: 17 (7%)
Debugging, Fixing Things, Recabling: 14 (5%)
Ordering Parts: 13 (5%)
Electronic Design: 12 (5%)
Soldering, Populating Boards, Wiring: 12 (5%)
Demos, Presentations, learning, etc: 9 (3%)
5: Spring Turkey is declared dead. Decided that a new robot will be built. Decided that
actuators will be based around Mike Wittigs prototype actuator he designed for undergrad
26: Meeting about DSP boards.
27: Called motor companies for catalogs.
4: Thought about leg configurations. Figured out actuator spec calculations.
5: Worked on robot configuration.
6: Worked on robot configuration.
12: Decided to go with bird-like structure with hips, (inverted) knees, and actuated ankles.
28: Simulations are working well.
29: Simulations. Hooked up Prototype actuator.
31: Tested Prototype actuator.
1: Tested Prototype actuator.
2: Prepared robot presentation.
3: Prepared presentation. Gave presentation.
4: Desired specs for actuators.
8: Ordered catalogs. CNC Mill Class. Tested Prototype Actuator.
9: Tested Prototype actuator.
10: Desired Specs for Actuators.
7: Meeting about DSP.
24: Played with DSP systems.
26: Played with DSP.
27: Played with DSP.
28: Analysis of force control in actuators.
29: Started learning Pro Engineer.
31: Learning Pro Engineer.
3: Learning Pro Engineer.
4: Learning Pro Engineer.
5: Learning Pro Engineer.
7: Learning Pro Engineer.
10: Hooked up Copley Amp, DSP.
11: Hooked up analog controller, Prototype actuator.
12: Worked on actuator controller.
13: Put Astro-Flight motor into Pro-E.
14: Designed fix for Prototype Actuator.
17: Ordered Parts for Actuator fix; machined parts.
18: Designed pulleys and tubes, started designing knee joint.
19: Worked on leg designs.
20: Machined parts for actuator fix; got parts and put it together; got it working and played with controller.
21: Found carbon fiber tube and adhesive companies. Figured out cable tensioning scheme.
22: Leg design.
23: Leg design.
24: Ordered parts: bearings, shafts, spacers. Ankle design.
25: Worked on DSPs.
26: Machining prototype knee joint.
27: Machining prototype knee joint.
28: DSP. Prototype knee joint.
9: Designed joints.
10: Actuator Design.
11: Actuator Controller.
12: Actuator Controller.
13: Actuator Controller.
15: Actuator Design.
16: Designed actuator shell, motor holder, bearing holder.
17: Machined actuator shell.
18: Actuator Design.
19: Designed new actuator shell and bearing holder. Ordered parts: bearings, springs.
20: Assembled and tested new actuator with new shell.
21: Started designing body frame.
22: Actuator Design.
23: Designed actuator pulley holder and tube clamp.
24: Got Quote for ball screws, carbon tubes. Machined cable holder part.
25: Designed Sensor stuff.
26: Body plate design. Made machining drawings for joints, body plates.
30: Body plate design.
31: Thought about boom attachments and stiffening the body plates.
1: Designed Reinforcement plates. Revised body plates. Made assembly drawings.
2: Made machining drawings of body plates.
3: Made body plate mock up. Realized it can't be machined on a Mill.
12: Faxed quote requests for carbon tubes and aluminum.
13: Ordered aluminum. Talked to people about using a water jet to cut body plates.
14: Cut out body plate on water jet. Ordered carbon fiber tubes. Started machining pulleys.
15: Machined pulleys.
16: Machined actuator angular contact bearing holders.
17: Machined actuator motor mounts.
20: Cut out body plates on water jet.
21: Cut out reinforcing plates on water jet. Figured out plastic injection molder "quick-shooter".
22: Machined 1 actuator shell. Started assembling an actuator.
23: Worked on actuator.
26: Machined actuator pulley holders.
27: Glued parts for an actuator. Started machining actuator shells.
28: Machined actuator shells.
29: Machined actuator shells.
30: Machined actuator shells.
31: Machined actuator spacers.
1: Machined reinforcer plates. Started putting body together.
2: Machined body separator parts. Put robot frame together. Redesigned sensor insulator parts.
3: Figured out injection molder. Redesigned part and mold. Made plastic sensor insulator parts.
4: Made drawing revisions for ball screws.
5: Called part companies for pricing and availability on motors, ball screws, mechanical parts, sensors, springs.
6: Ordered parts.
10: Installed Tango.
11: Started machining joints.
16: Machining joints.
18: Machining joints.
19: Machining joints.
25: Machined sample parts and practiced gluing carbon fiber to aluminum.
26: Tested glued parts. Revised some part drawings.
27: Cleaned up part drawings and faxed them off for quotes.30: Finished machining joints.
1: Finished up joint parts.
3: Assembled hip and knee joint sections. Designed hip spacer parts. Fixed assembly drawings.
4: Started boom design.
(In Japan for IROS '96)
12: Glued carbon fiber tubes to aluminum joints.
13: Hooked up an actuator. Started gluing parts for 2 more actuators.
15: Glued some actuator sensor parts.
16: Started machining hip bearing mounts.
18: Finished machining hip bearing mounts. Assembled hip joint and hooked up an actuator.
19: Started assembling another actuator.
30: Assembled an actuator. Assembled other hip joint.
31: Started mass assembling actuators.
1: Glued actuator stuff. Designed shafts for joints.
2: Analyzed actuators. Started designing force controller electronics.
3: Ordered parts: amplifiers, wire, connectors.
10: Figured out wiring.
13: Started designing force controller circuit.
14: Laying out Force Controller PCB.
15: Force Controller PCB.
16: Finished Force Controller PCB. Sent out board for fabrication.
17: Implemented an analog position controller and tested on hip. Started designing a boom.
19: Boom Design.
20: Boom Design. Figured out parts to order.
21: Ordered boom parts. Finished design.
22: Ordered tubes for boom. Figured out how to cable joints.
24: Started populating force controller boards.
27: Tested and debugged a force controller board.
28: Tuned force controller boards. Started mass soldering rest of force controller boards.
30: Finished soldering force controller boards, tested them. Machined shaft clamps.
31: Machined flattened shafts, pot holders, shaft clamps.
3: Reassembled hip joints. Computed and measured torque-voltage constants for
4: Started designing DSP/Filter break out board.
5: Designing DSP/Filter break out board.
6: Ordered parts for DSP/Filter break out board.
7: Designed and layed out power card. Revised boom design.
9: Started routing break out board.
10: Started working on DSP development software.
11: DSP development software.
12: Sent out break out board, power board, amp break out board for fabrication.
13: Assembled actuators.
14: Assembled actuators.
16: Machined boom swivel part.
17: Finished assembling actuators.
18: Started wiring actuators.
19: Ordered parts, and wire. Machining boom parts. Wiring actuators.
20: Finished wiring actuators. Tested them all.
21: DSP development software.
24: Made position pot holders. Ordered parts. Drilled holes in floor for boom.
25: Put the robot on ropes. Cabled hip and knee joints.
26: Designed and built DSP holder. Glued boom together.
27: Made cable tensioning screws.
28: Machined feet. Assembled feet. Worked on cabling.
3: Wired robot.
4: Wired robot. Soldered power boards and amp break out boards.
5: Wired power break out cables. Started soldering DSP break out board.
6: Soldered DSP break out board.
7: Tested and debugged DSP break out board.
8: DSP development software.
10: Ordered electronic parts. DSP development software.
11: DSP development software.
12: DSP development software.
13: DSP development software.
14: DSP development software.
17: DSP development software.
18: DSP development software.
19: DSP development software.
20: DSP development software.
21: DSP development software.
23: DSP development software.
24: DSP development software.
25: DSP development software.
1: Hooked DSP to robot.
2: Got DSP to talk to force controller boards.
3: Started wiring joint pots to DSP.
4: Finished wiring joint pots.
7: Made a kill switch. Wrote code for standing, calibration.
8: Worked on calibration, low pass filtering signals, torque coupling calculations, ramping on and off at start up and shut down. Got robot to stand on two feet!!
9: Got robot to do knee bends on two legs.
10: Debugged power supply drift - caused by large current loads of DSP.
11: Debugged hip torque error.
14: Implemented and tuned Virtual Knee Caps. DSP development software.
15: Stood (but not balanced) on 1 leg. Burnt out a motor and broke 3 cables.
17: Recabled with flexible cable.
18: Ordered motors, connectors.
29: Debugged noise problem, made new power cable.
30: Fixed an actuator sensor. Fixed a couple cables which broke.
1: Thought about boom counterweight. Broke and recabled cables.
13: Redesigned feet with rollers on them. Made boom counterweight. Got to stand on two legs and one leg (without balancing).
14: Tied wrangling rope. Worked on Virtual Toe Point algorithm.
15: Made long serial cable. Virtual toe point algorithm. Robot took half a step!!!
16: Made shorter feet and put them on. Tried to get robot to walk but had problems with free-wheeling ball screws.
18: Got robot to walk a little more smoothly but had to control x manually.
19: Worked on walking algorithm. Debugged actuator sensors: needed more spacing.
20: Changed out an actuator which was free-wheeling. Got robot to take several consecutive non-aided steps!!!!
21: Tuned walking algorithm.
22: Tuned walking algorithm.
23: Tuned walking algorithm. Double support math.
24: Double support math.
25: Double support math.
26: Debugged left hip velocity error.
27: Tuned walking algorithm.
28: Got robot to walk a full lap with no human intervention!!!!!
23: Fixed an actuator. Swapped in two new ball screws with tighter balls but still
24: Tried to debug ball screw problem.
25: Two cables broke. Recabled. Tried to get around ball screw problem but couldn't.
26: Worked on swing leg algorithm.
3: Got pre-loaded ball screws. Changed out two ball screws on actuators. This fixed the
6: Recabled and got to walk with assistance.
7: A couple cables broke. Recabled. Designed parts to fix cable breaking problem.
8: Trying to tune for faster walking.
9: Figured out algorithm for self-stabilizing speed.
10: Self-stabilizing speed algorithm.
11: Self-stabilizing speed algorithm. Couple cables broke. Recabled.
14: Got robot to take laps using self-stabilizing speed control.
15: Machined and installed a real knee cap.
16: Machined and installed other knee cap. Tuned Parameters.
17: Tuned parameters.
18: Tuned. Recabled. Debugged algorithm.
21: Installed parts to fix cable breaking problem. Helped but they still break. Tuned.
22: Got robot to walk continuously again.
23: Passive height control by using knee caps.
24: Lap in under 28 seconds!!! (Matching speed of Spring Turkey)
25: Couple cables broke. Recabled. Gave a couple demos.
28: Walking algorithm.
29: Walking algorithm.
30: Walking algorithm.
31: Walking algorithm.
1: Walking algorithm.
5: Walking algorithm.
6: Implemented virtual foot switch.
7: Walking algorithm.
8: Standing and knee bends with feet together and balancing.
14: Filmed for German TV film crew. Robot walked unassisted for almost an hour (3 laps at a time) without breaking a cable or falling!!! Did over 30 laps.
20: Robot walks up and down small ramp with no change in algorithm.
© Copyright MIT Leg Laboratory. All Rights Reserved.