Home

Quantum Autonomous
Navigation & Intelligence

QANTIS is a hardware-validated quantum platform for POMDP planning and multi-target data association. Validated across 45 experiments on IBM Heron QPUs, it demonstrates quadratic speedup in belief conditioning and solves NP-hard tracking problems via QAOA.

QANTIS v1.0
hardware-validated

We introduce QANTIS (Quantum Autonomous Navigation, Tracking & Intelligence System), a hardware-validated quantum platform that addresses autonomous navigation under uncertainty using quantum methods. The framework targets two core challenges: POMDP planning under partial observability and NP-hard multi-target data association in tracking scenarios.

For POMDP planning, belief conditioning traditionally costs O(P(e)-1). QANTIS leverages quantum amplitude amplification through a Grover-based belief oracle to reduce this to O(P(e)-1/2), achieving a quadratic speedup. A single Grover iterate amplifies observation probability from 0.179 to 0.907 — a 5.1x improvement validated on real hardware. The first closed-loop hybrid quantum-classical Tiger POMDP is demonstrated on superconducting hardware over T=8 decision steps with a maximum Hellinger distance of just 0.0149.

For multi-target data association (MTDA), the NP-hard assignment problem is formulated as a QUBO and solved via QAOA with fixed-parameter circuits. Hardware experiments across 45 runs on three IBM Heron QPUs (ibm_torino, ibm_fez, ibm_marrakesh) establish NISQ feasibility boundaries: ZNE is beneficial below ~100 ISA gates and harmful above ~1000, while FPC-QAOA produces meaningful results at up to 15 QUBO variables.

Quantum amplitude amplification reduces belief conditioning cost from O(P(e)⁻¹) to O(P(e)⁻¹˲), quadratically accelerating observation updates in partially observable environments.

First closed-loop hybrid quantum-classical Tiger POMDP executed on superconducting hardware (T=8 steps, maximum Hellinger distance 0.0149), proving real-time quantum decision-making viability.

NP-hard multi-target data association is cast as a QUBO and solved via QAOA with fixed-parameter circuits. Meaningful results demonstrated at up to 15 QUBO variables on NISQ hardware.

Systematic NISQ feasibility boundary established: ZNE beneficial below ~100 ISA gates, harmful above ~1000. Composable mitigation pipeline validated across 45 experiments on 3 IBM Heron backends.

45 experiments executed across three IBM Heron backends — ibm_torino, ibm_fez, and ibm_marrakesh — with composable error mitigation.

5.1x via single Grover iterate
quantum vs ideal distribution
post-mitigation recovery
IBM Heron backends

QANTIS is structured as three composable Python packages — shared primitives, POMDP planning, and multi-hypothesis tracking — all released under Apache 2.0.

Shared utilities, circuit primitives, error mitigation (ZNE, Pauli twirling), and backend abstraction layer for IBM Qiskit Runtime.

POMDP belief-state oracle construction, Grover amplitude amplification, closed-loop hybrid planning loop, and Tiger POMDP reference implementation.

Multi-target data association via QUBO formulation, FPC-QAOA solver, classical MHT baseline, and cost-matrix construction for tracking scenarios.

Bayram Yüksel Eker
Suayb S. Arslan
Özgür Nazlı
Mustafa Serhat Demirgil
Furkan Deligöz
2603.00785v1
February 28, 2026
31 pages, 4 figures, 12 tables
Apache 2.0
github.com/neuraparse/qantis
ibm_torino
ibm_fez
ibm_marrakesh
ZNE beneficial below ~100 ISA gates
ZNE harmful above ~1000 ISA gates
FPC-QAOA meaningful at 15 QUBO variables

QANTIS is open-source under Apache 2.0. Whether you're exploring quantum planning, NISQ error mitigation, or multi-target tracking — contributions and collaborations are welcome.