Place four chess pieces on assigned squares using both arms (ALOHA 2)

An ALOHA 2 bimanual rig: two ViperX 300 arms (6 joints each, with parallel-jaw grippers that open to about 7 cm) are mounted side by side at the back edge of a 1.6 m x 0.94 m desk, both reaching forward toward you. The "left" arm is the one at negative x (on the left in the front camera view), the "right" arm the one at positive x. This is a MuJoCo physics simulation with the MuJoCo Menagerie robot models: every object is a free rigid body that moves only through contact and friction (nothing is attached or teleported), so a loose grip, a fast swing or a collision can drop or knock things over. World frame in metres: +x to the right, +y toward the robots (the back of the desk), +z up; the desk top is z = 0.

Goal: Four chess pieces start in the two bowls: a light pawn and rook in the left bowl, a dark pawn and rook in the right bowl. Every piece has its own target square outlined in gold on the chessboard (<piece>_target). Place all four upright on their squares, using both arms: each arm must place at least one piece.

Success: Every piece's base centre is within 12 mm of its own target, rests on the board (within 4 mm of z = 0.0102), is tilted at most 10 degrees, released and at rest; and the pieces were placed by both arms (the arm that last held each piece counts). Success is judged by the episode server from the physical state after you call robo done and the robots have held still for about 10 steps (0.5 s). "Released" means neither arm touches the object; "at rest" means it moves slower than 1 cm/s.

Goal fields in robo observe: <piece>_target for each piece and square_size (0.04 m). Pieces: base disc 2.6 cm across, a 1.8 cm stem, pawn 6 cm tall, rook 6.1 cm.

Controls. Both arms move at the same time on one clock. robo act L_DX L_DY L_DZ L_GRIP R_DX R_DY R_DZ R_GRIP gives each arm its own command in the same step: DX, DY, DZ move that arm's commanded gripper target by 2 cm per unit along world x, y, z (each in [-1, 1], so 0.25 = 5 mm); GRIP 0 keeps the fingers as they are, any positive value closes them, and a negative value -f opens them to fraction f of full width (-1 fully open). One step is 50 ms. robo move-to and robo grip are not available in this task, and the DX DY DZ GRIP form in the general instructions below does not apply: every robo act needs all 8 numbers. Inverse kinematics turns each target into joint commands; the grippers always point straight down.

Observation. robo observe reports, for each arm (prefixed left_ / right_), hand_pos (the measured point between the fingertips), hand_target (where the arm is being driven; it can run up to 5 cm ahead of the hand, which then catches up), gripper_open (0 closed .. 1 fully open) and touching (task objects in contact with that arm); for every task object <name>_pos (the centre of the object's base, where it touches the surface below), <name>_tilt_deg (tilt from upright) and <name>_yaw_deg; plus the goal fields listed above. robo observe --image saves a picture from the front camera (--camera overhead looks straight down with +y at the top of the picture; --camera overview is a far corner view).

The step budget is 2000 steps (100 s of simulated time).