Cyber Spider 
(all content on one page)

   
A cheap 8-legged motorized spider for around $20. I got this on sale from ToysRus. It is called "Cyber Spider" and is made by Wow-Wee inc. Eventually, this toy will become obsolete but by examining and understanding the mechanics in this toy I can utilize those ideas in other designs. I'm specifically looking for forward,  reverse and turning capability.
   
The body is about the size of a hand. It came with an infrared controller seen here on the arm. The spider could move forward, turn right slowly and turn left slowly. It also had sound effects, LEDs and a light sensor.
   
Removing the top reveals a gearbox with dual motors and a set of printed circuit boards. I'll toss the boards since I want to focus on understanding the mechanical design. Later on, I could add my own microcontroller brains and sensors for autonomous operation.  
   
A close-up of the dual motors. The motors have a worm gear on the end of their shaft. The left motor controls the sweeping of 4 legs on the left side and the right motor controls the sweeping of 4 legs on the right side.
   
Each motor drives 4 round cams located on the right and left side of  the spider (cams are painted red). There are protrusion that come out of each cam. It reminds me of the handle on a crank. I put a white dot on the end of each cam "handle".  You can clearly see the roughly 180 degree phase relationship between the cam handles. 
   
The cam movements are linked together on each side through a set of in-line gears.  
   
Here's a close-up of the gearbox separated from the spider body . 
   
The 4 white posts on the spider base are the mounting points for the legs.
   
The leg has a joint with 2 degrees of freedom. One degree of freedom is formed from the hollow white swing post seen here between my fingers. This post swings about the white pivoting point on the spider base (seen in the previous picture). The movement about this axis allows the spider leg to swing forward and backward..
   
The second degree of freedom allows the spider leg to lift and lower. The end of the leg body painted in red pivots about the white swing post.
   
This shows the white swing post centered about the end of the red leg joint.
   
This shows the white swing post leaning to the left of the red leg joint.
   
This shows the white swing post leaning to the right of the red leg joint.
   
Picture showing how the leg joint mates with the cam handle.
   
A shot showing the white swing post on the leg joint and how it will line up with the white pivoting point on the spider base.
   
The leg movements are offset from each other according to where they mate with the cams.When the cams turn they form a sweeping motion that lifts the legs, moves the legs horizontally and then lowers the legs.  Imagine the legs acting like oars pivoting on the attachment points of a row boat. The sweeping circular motion you would perform with your arm is being done by the rotating cams. 
There are always enough legs on the ground to make the spider statically stable. The legs on each side have roughly a 180 degree phase difference. Ideally, legs across from each other should be 180 degrees off in phase for the best balance but it's not critical since the robot will still remain stable. This phasing of two legs always in contact with the surface on each side of the robot reminds me of a tripod gait but with an extra leg. I'm going to call this a quadpod gait. 

The stock robot movements were very limited since the motors could move in only one direction and at two speeds.  By being able to reverse the motors, the spider could move backwards and perform a sharp turn.

The spider control is similar to a tank control. Forward motion is performed by sweeping legs on each side in a forward direction. Sweeping legs on each side in the reverse direction causes the robot to move backwards. A sharp turn is performed by sweep the legs on one side forward and the other side backwards. A slow turn is performed by sweeping the legs on each side in the same direction but slowing down the sweep speed on only one side. The bug turns in the direction of the slower sweep. Movements were good on flat surfaces but the spider had problems on carpet because the legs don't move high enough and end up dragging against the carpet.

In my tests the motors have two speeds created by varying the number of battery cells connected to the motor terminals. 4 AA batteries connected to a motor provides fast speed and 2AA batteries provides slow speed. The spider has a battery compartment underneath for 4 AAA cells. I had a problem getting the cells to come into good electrical contact with the battery terminals. It seems like the plastic molding in the battery compartment was out of tolerance (welcome to the world of cheap molding and plastic;). In the videos, tests were performed using battery cells located externally. This did make the robot lighter so the movements may not be as good when cells are onboard.

Overall, the mechanics are flexible enough for a wide range of movements if the motors are made reversible. The leg joints, cams and gearbox provided a lot of insights into the minimum mechanisms needed to sweep each leg. I think of this design as a redesign for manufacturability of my more sophisticated custom built robot bug.  In fact, by changing the phase relationship between the legs other gaits could be formed with the Cyber Spider.