Skip to content

Alif Jakir

Omnidisciplinary inventor

philosopher–scientist–futurist

Hi, I’m Alif—welcome to my site. My work sits at the intersection of artificial and organic intelligence: how minds arise in neural and machine substrates, and how we can design systems that amplify critical thinking and partnership with humans rather than substituting for them. The full layout, contact links, and research sections load with the interactive site.

Loading a bunch of web magic

Fonts are traveling; typography will arrive shortly.

MosaicDrone

MOSAIC — Modular Object-Oriented Stackable Autonomous Intelligent Craft

MosaicDrone is what I've been building toward since the modular drone work at TARS: programmable aerial voxels — a you can sculpt with gestures. Individual units dock to form larger structures, share power and data through standardized connectors, and use autonomous recharge rotation to keep formations . Think of it as “Air Legos” for research, art, and adaptive infrastructure. The name MOSAIC spells out the design philosophy.

What MOSAIC means

Each letter maps to a core design principle:

  • Modular — Standardized mechanical, electrical, and software interfaces across units and payloads so any part can be swapped or upgraded.
  • Omnidirectional — True 6-DOF control (decoupled translation and rotation) for precise assembly in any orientation, not just horizontal flight.
  • Self-Assembling — Reliable mid-air docking, structure growth and shrink, and automated reconfiguration so the swarm can form and reform shapes.
  • Autonomous — Distributed planning, fault tolerance, and continuous operation without a single central bottleneck.
  • Intelligent — Learning-ready stack for perception, scheduling, and human–swarm interaction (e.g. gesture and AR).
  • Craft / Collective — Each unit is a craft; once docked, they behave as one meta-craft with a unified actuator set and control allocation.

How MARP inspired MosaicDrone

MosaicDrone didn't start from zero. It grew directly out of the Modular Aerial Robotics Project (MARP) at TARS Lab, Clarkson University — the same effort that produced our designs, the v1 pancake and v2–v3 box drones, and the footage on the MARP page. That work is the direct precursor in both concept and hardware.

At TARS we proved that modular, stackable airframes were feasible. The v1 pancake showed you could stack units on top of each other into a conic formation. Then v2, the big heavy box with ball-and-socket joints at the corners, and v3, the small box with magnets on the ends. We explored VR teleoperation, biometric pilot ID, and swarm coordination. The design draft spelled out the truss structure, universal connectors, heterogeneous stacking, and the path from single units to formations. Crazyflie came into the pipeline to iterate fast on stacking logic while custom builds (Pi, Navio2, Jetson) carried the full hardware stack.

Those experiments answered a central question: can small aerial units physically connect in mid-air and behave as a larger system? Once we had evidence that they could — and that the right interfaces (mechanical mating, power, data) were the key — the next step was turning that into a general-purpose voxel: omnidirectional so it can dock in any orientation, sharing power and data through the dock, reconfiguring the collective's control allocation on attach or detach. MARP gave me the stacking paradigm and the early hardware; MosaicDrone is the evolution into a full MOSAIC stack — MOMAV-style actuation for 6-DOF, spec'd docking with pogo pins and state machines, swarm orchestration, and gesture-driven formation control. Same thread, carried forward.

Technical foundation

The research basis is MOMAV (Marco's Omnidirectional Micro Aerial Vehicle) — a fully actuated multirotor with rotating arms in a symmetric 3D geometry (e.g. octahedral) that decouples translation and rotation. I'm adapting that idea for modular swarms and docking. Arms align to the vertices of a 3D solid; rotating them with slip-rings and modified servos gives precise 6-DOF control. An SQP-based control allocator tunes throttle and arm velocities against objectives and constraints; when drones dock, the actuator set becomes the union of attached modules and the mixer updates online so the aggregate is controlled as one object.

Mid-air docking is where the hard engineering lives: alignment geometry (chamfered cones/pins), magnet-assisted latch, pogo-pin arrays for DC bus and CAN or Ethernet. The state machine goes Seeking → Aligning → SoftContact → Latching → Verify → Bonded, with emergency release within 200 ms. Power and data handover are brownout-safe (pre-charge, inrush limiting). Perception uses MoCap/UWB indoors and GNSS RTK + VIO outdoors; docking perception (visual/IR beacons) feeds the state machine.

The swarm layer runs on ROS 2: formation planning, role assignment, collision envelopes, recharge scheduling so units peel off to charge and rejoin without losing the formation. Human input — gestures, AR, or app — maps to high-level formation edits with sub-80 ms latency targets. Acoustic design (larger slower props, ESC PWM at 48–96 kHz, structural damping) targets low-noise operation for indoor and human-in-the-loop use.

What I'm building toward

  • Omnidirectional unit flight via MOMAV-style actuation and control allocation.
  • Mid-air docking with orientation-agnostic connectors (mechanical + power + data).
  • Aggregate control that reconfigures mixers when drones attach or detach (formation as one object).
  • Low-noise operation through prop design, ESC tuning, and structural damping.
  • Human-in-the-loop interaction via gestures/AR with sub-80 ms end-to-end latency.
  • Persistent uptime via autonomous recharge/battery-swap scheduling and voxel substitution.

Applications

As a robotic research platform, it's a testbed for fully actuated flight, docking, allocation, and swarm controls. For adaptive structures, the swarm becomes reconfigurable sensor arrays, antenna arrays, or temporary installations. For human–robot interaction, it's gesture-controlled formations with low latency. The repo and economics docs also explore recyclofacturing and landfill mining: mobile cells (e.g. WAAM, 3D printing) working with drone swarms for material handling and processing.

Roadmap

  1. MVP indoor lattice with 4–8 units (MoCap/UWB, manual battery swap).
  2. Reliable mid-air docking + pogo power/data and safety envelopes.
  3. Dynamic mixer reconfiguration for bonded aggregates.
  4. Recharge perches and autonomous rotation scheduling.
  5. Low-noise blades and ESC tuning package.
  6. Outdoor pilot with RTK + VIO and safe distances.
  7. Open choreography/gesture API and AR overlays.

Repository and documentation

Everything lives in a dedicated repo with full technical specs, subsystem status, safety and operations docs, hardware CAD and BOM, simulations (e.g. Isaac Lab), and economics.