Mechanical Sub-system

April 27, 2013

Final CAD Shell Design and Final Shell

We 3-D printed the final shell based on the CAD model. The CAD design and the shell are shown below, respectively.

April 8, 2013

Stabilizer bars added to chassis

Since the standoffs in the chassis were free to rotate, we found that they needed to be held in place or the screws on the other sides of the sidewalls would loosen. A solution to this problem was discovered when one of the sidewalls snapped due to lateral force and the need to strengthen the chassis became apparent. Thus, stabilizer bars were designed and printed from the 3D printer. The "bar", which is actually a flat piece, connects the two upper-most standoffs and features hexagonal holes that keep the standoffs from rotating. A stabilizer can be seen in the figure below.

March 14, 2013

New Bean Shell implements "egg" geometry study

A new shell, now often referred to as "Bean" or "Magic Bean", has been started. Compared with the old cage-like design, its length and width dimensions are swapped meaning that the front and back edges extend farther than the sides. Its shape is partially elliptical, but flat on the bottom for reducing unnecessary mass and allowing for proper self-righting and future rotation. Structurally, it is solid and built of only three large pieces, but it is very thin and some experiments will need to be performed to determine whether this approach is viable. The figure below shows a completed side section of the Bean shell.

March 2, 2013

Mockup for adjusting the shape of the second prototype shell design

We tried a new shape for the second prototype of the shell to address the issue of the spring sticking out the cage. We first 3D-printed a scaled-down egg-shaped  mockup, which is shown in the figure below. In the picture, the mockup is upside-down with box of screws simulating the mass of the inner chassis which offsets the center of mass.

The design of  the second prototype shell with the  adjusted shape

After successfully validating the idea, we worked on the CAD design of the second prototype shell to readjust the shape, as shown below.

February 14, 2013

First prototype shell

Today we demonstrated the first prototype shell which is able of doing orientation recovery in  all directions. This means that it should enable the robot to fall back to the standard launch  position after landing on its side, upside-down, or any angle in between.

The figure below shows the completed new spherical shell combined with parts of the rotation and jumping mechanism . Its overall shape consists of two hemispheres connected by a cylindrical section in  between. This shape is similar to a rounded pill. The cylindrical section in the middle serves two  purposes: first, it works more naturally with the fiberglass strips the bend upward through the  bottom of the enclosure. And second, it constrains the robot to rolling in the forward-backward  plane. We don’t want the robot to remain on its side after landing.

It is worth noting in this design, we moved up the sidewalls relative to the shell in order to solve the issue of the spring sticking out the cage when charged, but this was not the best approach because we lost some height for charging the springs.

CAD design of the second prototype shell

The first prototype shell validated the idea, but certain portion of the shell was fragile to impact and the whole shell was heavy. As a result, we went on designing the second prototype shell. The two figures below are the CAD designs.

February 6, 2013

Self-righting and rotation mockup

Further work was done on fabricating the self-righting/rotation mockup for the current prototype - likely the last additions we will be making to it before moving on to a second prototype that incorporates all that we have learned thus far.

The original idea we had (seen in previous post) is that the robot would rotate on its side using just a corner of one of the wheels. So the robot would rest on one wheel and its foot with the leg fully extended. Then, it would roll into the pre-jump position with the leg compressed, since the center of gravity is biased toward one edge of the circle.

This turned out to be problematic when we realized that the rotation orientation and jumping orientation are orthogonal to each other. The robot's present capability is essentially determined by how it landed after the previous jump (i.e. on its side or straight).

To fix the issue, we decided to move the wheels toward the forward end of the base and test a dual-hemispherical (see Figure) shell that will force the robot into only a straight orientation. To do this, we took the double ring design that can be seen in the previous post and added outer rings extending to the sides to give the shell its spherical shape.

Figure: Front and side views of the proposed design that illustrate the robot's orientation when rotating and jumping

January 31, 2013

First attempt on designing the self-righting mechanism

Figure below shows the current robot structure which proves the concept of rolling back to the normal position in one dimension with mock wheel attached to the bottom of the spherical structure. We also migrated from the single-spring jumping mechanism to a dual-spring one.

Figure : Integrated system – Jumping mechanism with mock self-righting mechanism and rotation mechanism

Although we validated the idea, we still had two issues with this model. The first issue was that since the supporting chassis was not designed for a dual-spring jumping mechanism, there was friction between the spring-charging cable and the motor, which consumed a lot of energy; the second issue was when charged up, the spring would stick out of the rings, which sometimes prevented the robot from rolling back to the desired attitude.

January 24, 2013

Improved jumping mechanism

The video below shows the the jumping of RooBot after the following improvements were made on the design demonstrated during the validation experiment in Fall '12:

1. Reduce friction on the shaft of the spring clutch mechanism

2. Improve spring clutch release mechanism (servo actuation was replaced by passive releasing)

3. Add a better traction base to the jumping chassis

4. Reduce weight (temporary the battery and electronics were removed to compensate for the estimate weight for the next design)

Brainstorming conceptual designs for self-righting and rotation mechanism 

After discussions with Dr. Ben Brown, it became very clear to have a design for self-righting mechanism before designing the rotation mechanism. The team conducted numerous brainstorming sessions for both mechanisms. Figure 1 shows models of a few self-righting mechanisms as an outcome of those sessions. The model in figure 1(a) shows a boat-shaped structure with four legs (two on each side). The boat shape made sure that the robot always falls along one of the sides of the boat and the legs that are attached through strings will pull up through the chassis to make it upright as the jumping springs compress and move inside the boat-shaped structure. But the system demanded a passive mechanism as active mechanisms will add to size and weight of the system. The model in figure 1(b) shows two rotating conical wheels of large height around the main chassis of the jumping system. Such a design will make sure that the structure never falls down on its sides and having the center of mass of the robot below the center of circular base of the cones will make sure that robot stands upright as the jumping springs compress and move inside the conical wheels. These conical wheels can also assist in rotation. The third model shown in figure 1(c) has two fixed hemispherical wheels with flat sides that are mounted on the sidewalls. It comes upright using the same concept as the model with conical wheels. The structural advantage of the curved surface in the hemispherical model helps reduce the size of the robot maintains a symmetrical design.

Figure 1: Different configurations of self-righting structure

November 19, 2012

Mechanical side of the pre-prototype

Today we demonstrated the "pre-prototype" design of the RooBot.  This included wirelessly commanding the RooBot over Xbee to start winding the fiberglass spring and subsequently release the clutch.  This design represents a major milestone in our development and will inform our final prototype set to be completed for our team's fall deliverable on December 3. The Figure below shows the mechanical design of the "pre-prototype".

November 7, 2012

CAD and fabrication of the pre-prototype

Today we demonstrated the completed Solidworks model of our Pre-Prototype showing the implementation of the spring clutch and fiberglass spring.  In addition, physical geometries of all other relevant components were also detailed in the model including the battery, primastic joint spring mounting, chassis, drive motor, pulley, and servo as shown in the figure below.

Solidworks Model of Pre-Prototype

Fabrication was also started in earnest and the only remaining components that remain to be built and/or sourced are the brackets holding the motor and spring assembly.  The figure below shows the manufactured sidewalls (made of .25" Delrin plastic), standoffs, spring clutch, bearing supports, and the spring clutch side bevel gear.

Pre-Prototype - Partially fabricated

October 24, 2012

Spring launching experiment

Today we conducted a demonstration of spring launching capability. Below is our test stand clamped to our lab bench. 

Fiberglass Spring test stand

The springs were fixtured using L brackets to sandwich the fiberglass at each end and distribute the loading condition across the material.  This prevented splitting across the primary axis.

Fiberglass Spring Fixturing

Springs were both loaded and launched. At this time, we are primarily concerned with force versus displacement for various spring sizes so that we can understand the energy properties and thus inform our form factor.

The following shows preliminary data of eccentric force versus vertical displacement.

Eccentric Loading versus Displacement for .030" springs

While more exact methods are required to measure height, our best launch for a 1.5" wide, 4 inch long, .030" thick spring with payload of 25g was approximately 40".  We are on our way! More results to follow!

October 23, 2012

Fabrication of the test springs

The team also got into the RI machine shop to manufacture test springs.  4" long strips of various thicknesses and widths were produced.

Marshall and Mike measure and cut test springs

October 18, 2012

Test stand for spring testing

Our team is developing a fixture to experimentally compare different fiber glass springs.  The figure below shows the test stand that has been manufactured to date.  

Testing apparatus for comparing different fiber glass springs