AirCanvas: A Multimodal Framework for Spatial Mathematical Reasoning
Technical Brief on Hybrid Edge-Cloud Perception, Kinematic Smoothing, and Embodied STEM Cognition
// Authors & Affiliations
Chirag Nahata
Research Associate, Quantum Labs
Engineering Staff at Quantum Tiger specializing in core AI systems and foundation model engineering.
LinkedIn Profile →
Snigdha Ghosh
Research Associate, Quantum Labs
Engineering Staff at Quantum Tiger focusing on research-driven AI and multimodal interfaces.
LinkedIn Profile →
Somyadip Ghosh
Research Associate, Quantum Labs
Engineering Staff at Quantum Tiger contributing to applied AI engineering.
LinkedIn Profile →// Executive Abstract
Traditional computer interfaces impose a cognitive "Input Bottleneck" in STEM fields: typing mathematical formulas or chemical compounds disrupts thinking. AirCanvas introduces a spatial computing framework combining real-time computer vision (MediaPipe 21-point skeletal tracking) with Multimodal Large Language Models (MLLMs) to solve mathematical problems drawn via free-space hand gestures. The paper outlines the kinematic smoothing algorithms, finite-state gesture machines (>95% precision), and a thermodynamic split-compute architecture that keeps high-frame-rate tracking on 5–10W edge hardware while offloading deep visual chain-of-thought reasoning.
Research Methodology & Theoretical Framework
Prototyping and real-time kinematic evaluation using 30 FPS edge video feeds, Exponential Moving Average (EMA) tremor attenuation (8–12 Hz), finite state machines, and zero-shot visual chain-of-thought prompting on multimodal models.
Key Findings & Technical Breakthroughs
Overcoming the Symbol Grounding Problem: Free-space hand drawing restores embodied cognition, closing the feedback loop between physical motor action and abstract mathematical representation.
Split-Compute Energy Optimization: Running continuous visual tracking on edge CPU draws only 5–10W, preventing thermal throttling that occurs when running heavy Vision-QA models on local edge devices.
Kinematic Tremor Attenuation: Adaptive filtering resolves human physiological tremors without requiring expensive haptic input devices.
Accessibility Democratization: Provides an intuitive gross-motor interface for students with dysgraphia and motor impairments.
Policy & Industrial Implications
Paves the way for classroom-ready, sovereign educational technology and natural spatial interfaces for technical workers in India and emerging markets.
// Subject Classifications & Keywords
Related Quantum Tiger Research
Training vs Inference: Path to India’s AI Sovereignty
India’s enterprise AI adoption is mainstream (~87% enterprise adoption according to NASSCOM), but strategic sovereignty sits on the ability to train—not merely use or serve—large-scale models within national jurisdiction. This paper demonstrates that adoption does not equate to sovereignty. Without domestic foundation training capacity, model IP, training data, and strategic control remain offshore. We propose an actionable, funded 5-year program to establish 3–5 regional National AI Training Hubs, an IndiaAI Compute Credits facility, sectoral federated data trusts extending AI Kosh, and an Indian Sovereign Model Certification (ISM-Cert) framework.
Physics of Computation & Non-Equilibrium ThermodynamicsThermodynamic Constraints on Artificial General Intelligence
Artificial General Intelligence (AGI) is commonly discussed as an abstract algorithmic construct. This paper argues that AGI must instead be understood as a physical system governed fundamentally by the second law of thermodynamics. We demonstrate that learning, reasoning, memory updates, and parameter optimization are inherently irreversible processes with unavoidable physical entropy production. Using Landauer’s principle (Emin = kT ln 2), statistical mechanics, and non-equilibrium physics, we derive lower bounds on the energetic cost of intelligence and establish a fundamental heat dissipation ceiling (Qmax) that constrains infinite model scaling.