-
Design criteria and Desired Functionality:
+
Functionality:
+
+ The desired functionality starts with the user inputting their desired
+ move into the terminal and ends with the robot arm using the suction
+ cup gripper to do pick and place operations. The robot arm plays the
+ users move as well as a move of its own. To pick up a piece the UR-10
+ robot arm must move its end effector to the appropriate location,
+ activate a pull state the suction cup and lower slowly until a seal is
+ formed. If the initial attempt to pick up the piece fails, the end
+ effector moves toward the most probable location that a seal will be
+ formed. The robot arm then moves the piece to the appropriate location
+ switching between the suction states as needed. This continues until
+ their has been a check mate, or the user surrenders.
+
+
+
Design Goals:
+
+ The following is a list of our criteria needed for what we considered
+ to be a successful implementation of the UR-10 with the smart suction
+ cup gripper to play chess:
+
-
- The robot arm can pick up and place down a piece correctly as we
- desired.
+ Use computer vision to identify the center of the board and adjust
+ mapping of board accordingly.
-
- The robot arm need to have the ability to perform its own move base
- on the current game situation so it can play with humans.
+ The terminal accepts a user input and returns the next move played,
+ resulting in pick and place operation to move pieces according.
-
- Perform haptic search so it can pick up the chess pieces accurately
- even if we only have the rough idea of where the piece will be.
+ Upon an unsuccessful suction seal, the UR-10 move the end effector
+ towards the most likely location for a better seal.
-
- The robot arm need to detect and deal with the edge cases in chess
- such as upgrading and castling.
+ The suction cup should be appropriate state (Push, Pull, Off)
+ corresponding to the action or nonaction happening.
-
- The robot arm can recognize the current game using machine learing
- and computer vision so we don't need to record the game stats.
+ Consider chess edge cases. (i.e. Capturing, Castling, Promotion)
-
-
Design Choices and Details:
- Although the game we want the robot arm to perfrom is just a normal
- chess game, since we want to emphasize the idea of smart suction cup.
- We 3D print our own chess pieces so that it is thin and flat that will
- be hard for a normal grippers to pick up. We also make our own chess
- board by cutting a wooden board that each square of the chess will be
- 3 inch by 3 inch.
+ With this combination of design goals, the UR-10 robot arm should be
+ able to repeatability and reliably play countless number of chess
+ games. Additionally, it was important that we design our chess pieces
+ such that it demonstrates appropriate use of haptic feedback search
+ using the smart suction cup technology.
+
+
+
Design Decisions
+
Custom Chess Pieces:
+
+ Our team custom designed flat chess pieces in order to highlight the
+ application of haptic search. The pieces were designed using the
+ computer-aided design software OnShape, and are pictured below. When
+ designing these pieces we ensured there was ample surface area for the
+ gripper to form a seal, while adding still some degree of complexity.
+ The some pieces have sharp corners and holes making it difficult for
+ the haptic search to be successful. Further, we wanted the pieces to
+ be light to demonstrate the need for a push state. These pieces are
+ light enough such that when the section gripper is in the off state,
+ the piece remains attached to the gripper due to a residual vacuum
+ left within the system.
+
+
+
Hardware and Electronics:
+
+ We built off the system provided by Dr. Jungpyo Lee. Pneumatically,
+ the system consisted of a single solenoid to control the flow of air
+ to an in line vacuum generator. The solenoid was electronically wired
+ to a n-type MOSFET, whose gate pin was controlled by a ESP32
+ microcontroller. In it's initial state, the system had no means for
+ pushing air out of the suction cup end effector. Together with Dr
+ Jungpyo Lee, we devised a new system which included push. This newly
+ devised system saw the addition of 3 more solenoid-MOSFET pairs. The
+ updated pneumatic lines and circuit are shown in the image below,
+ accompanied by a schematic of the the circuit as well. Initially, we
+ experimented with splitting the line using only two solenoids. The
+ idea being the solenoids would be on separate lines and activate
+ corresponding to being in a push or pull state. This did not work as
+ intended air was escaping through the vacuum generator. We decided to
+ get the third solenoid to isolate the vacuum generator when going into
+ the push state.
+
Gameplay
+
+ We out sourced the logic behind the game play to an open-source chess
+ engine called Stockfish. Due to hardware limitations we were
+ restricted to using older version. We drew up a 24 in x 24 in chess
+ board, resulting in squares that were 3 in x 3 in. We decided to hard
+ program the location of each squares center, however this does not
+ grantee a piece will be in an orientation that can be picked up on the
+ first try. Additionally, we programmed in extra locations on the off
+ the board for each team to have a graveyard and place where promotion
+ pieces can be grabbed from. This way we could handle the common
+ capture and the less frequent edge cases like promoting and castling.
+
+
+ 
+ 
+ 
+ 