This video details the initiation of a new side project: a "Ball Butler/Primer/Launcher" robot designed to automate ball feeding for the speaker's main "Jugglebot" project and eventually for personal juggling practice. The project focuses on miniaturization and serving as a testbed for new engineering concepts.
The speaker outlines the multi-stage robotic arm design, featuring a string-driven linear hand, a pitch/tilt mechanism, and a yawing base. Key design elements include a custom ratchet-based string tensioning system, considerations for motor current limits in a compact form factor, and innovative 3D printing techniques to overcome assembly challenges. A significant part of the discussion revolves around exploring advanced control strategies, such as the use of absolute encoders for all axes (except the hand) to eliminate homing sequences, and the potential for output-side encoders to ensure precise kinematic accuracy. The speaker also discusses the future implementation of a mechanical ball-loading system from a hopper and the ambitious goal of making the robot battery-powered. Challenges like unreliable encoders and issues with motion capture marker ambiguity are also highlighted, alongside reflections on the CAD design process itself.
Summarized Transcript:
0:00:29 New Side Project Introduction: The speaker announces a new project: the "Ball Butler/Primer/Launcher" – a robot designed to throw juggling balls.
0:01:16 Project Goals: The immediate goal is to create a robot that sits on a shelf and feeds balls to "Jugglebot" as needed. The long-term goal is a portable robot for personal use in the park to eliminate bending over to pick up dropped juggling balls.
0:02:48 Project Justification: Despite Jugglebot not yet being at a stage where it constantly drops balls, manual reloading of the current five tracked balls is already tedious. This side project serves as a platform to experiment with new techniques.
0:04:20 Visual Branding: Future Ball Butler videos will feature a blue background in thumbnails to distinguish them from Jugglebot content.
0:06:01 3D Printed Mesh: The speaker demonstrates a flexible 3D printed TPU net that can be crumpled and springs back, and a more rigid PLA version. A custom OnShape feature script for "meshifying" objects is also shown.
0:09:16 AMS Filament Indicator: A small, self-made 3D printed slider is presented for tracking filament types loaded into a 3D printer's Automatic Material System (AMS).
The design is heavily inspired by Jugglebot's hand, using an aluminum extrusion and linear rail.
0:11:40 Tensioning: A modified tensioning mechanism for the string requires moving the entire motor assembly, which is a "necessary evil" due to space constraints and the need for fixed string ends to prevent slipping.
0:14:10 Miniaturization & Performance: Testing on Jugglebot showed the shorter hand (160mm stroke) could throw balls over 1 meter high by pushing motor current limits (100A soft, 120A hard), sufficient for Jugglebot feeding.
0:17:05 Overshooting & Testing Platform: Concerns about overshooting with high accelerations are noted, but the Ball Butler is intended as a test platform to experiment with more aggressive parameters without risking the main Jugglebot.
0:20:10 Encoder Slippage: Experience with encoders slipping on Jugglebot's leg shafts highlights the need for more robust solutions for fast-moving components.
0:20:52 Magnetic Encoders: Previous AliExpress-sourced magnetic encoders proved unreliable; considering a different version or utilizing the Odrive's onboard encoder.
0:23:00 Odrive Placement: Mounting the Odrive Pro directly on the hand motor is being considered for compactness, reducing complex cable routing.
0:23:38 Linear Encoder Idea: A viewer suggests using optical linear encoders (like those in calipers) for absolute position sensing, which the speaker finds intriguing for potential future Jugglebot linear actuators.
0:25:59 Minimal Electronics on Hand: The goal is to avoid mounting any electronics directly on the fast-moving hand to prevent cable issues.
This stage uses a string-driven capstan drive system (string fixed to the spool to prevent slipping).
0:28:40 Ratchet Tensioning: A robust 3D printed ratchet spool, tightened with a custom tool, provides secure string tensioning.
0:31:02 String-on-Plastic Friction: Potential for the string to cut into the plastic due to tension is a concern, as integrating pulleys proved too complex.
0:37:27 Torque Transfer: A central aluminum spool, manufactured by a sponsor, will transfer torque from the motor to the hand assembly, supported by bearings.
0:38:46 Assembly Compromise: Assembly requires drilling directly through the aluminum extrusion, a departure from previous practices.
0:40:01 CAD Best Practices (Tangent): The speaker shares a tip for OnShape users: use "configuration variables" for fastener libraries instead of "configurations" to dynamically set lengths and avoid manual updates and overly long bolts in models.
0:44:50 Printing Challenges (Main Ring): The main ring's geometry, with internal cones for bearing contact, makes single-piece 3D printing difficult without excessive supports.
0:45:54 Two-Part Printing Solution: Inspired by another functional print video, the ring will be printed in two parts and super-glued together to optimize printing orientation and minimize support.
0:49:15 Output-Side Encoders for Precision: The speaker discusses the idea of adding a second encoder on the actuated output of the pitch/tilt and yaw mechanisms (in addition to the motor encoder). This would provide absolute position feedback and compensate for any slack or inaccuracy in the string drive, ensuring precise control.
0:52:27 Absolute Encoders for Homing: For motions less than 360°, absolute encoders would eliminate the need for tedious and potentially unreliable homing sequences upon power-on.
This stage uses a central motor with belts for nearly 360° rotation.
0:56:16 Friction Reduction: Ultra-high molecular weight (UHMW) tape will be applied to contact surfaces to reduce friction for the belts.
0:57:03 Overall Homing Strategy: With absolute encoders on stage 2 and the base, only the hand would need homing (using a reliable current-spike detection method).
0:58:16 Remaining CAD Work: The base still requires significant design work for mounting and electronics integration.
0:58:39 Odrive V3.6 for Base/Tilt Motors: Older, dual-motor Odrive V3.6 boards will be used for the base and tilt motors to utilize existing hardware, potentially requiring an electronics chamber.
1:00:04 Proximity Sensors (Alternative Homing): Inductive proximity sensors are considered as a contact-free homing alternative to current-spike detection or absolute encoders.
1:01:49 CAD Design Process (Reflections): The speaker shares their common experience of initial "pit of despair" when facing complex geometry, overcome by starting with crude solutions and refining them, and considers doing more concept sketches.
1:04:41 Ball Loading Mechanism: A mechanical lever system, inspired by a 3D printer's nozzle cleaning process, is envisioned. The hand's downward motion would activate a scoop to pluck a single ball from a hopper into the hand, allowing for quick reloading.
1:13:52 Battery Power: The speaker plans to use batteries to power the robot, a new and somewhat daunting experience, as a valuable learning opportunity.
1:15:27 Motion Capture Marker Ambiguity: The speaker expresses frustration with designing retroreflective marker configurations for motion capture that remain unambiguous even when some markers are obscured or if the system tries to "flip" the object's orientation.
1:22:57 Project Timeline: No stream next week; the speaker plans to continue the Ball Butler project in two weeks, aiming to complete it by year-end, driven by the desire for a personal ball-feeding robot.
1:26:29 Inspiration Credit: The capstan drive for the pitch/tilt mechanism was inspired by Aad Musa's video.
1:28:06 Seeking Feedback: The speaker is open to suggestions, particularly regarding the hand's linear encoder, given the high speeds (max 5 m/s, very fast accelerations).