Positive Velocity Feedback

• Positive velocity feedback is a mechanism where a system’s current or delayed velocity is amplified via a positive gain to reinforce motion.
• It employs mathematical models that integrate direct or delayed velocity terms to enhance responsiveness while posing risks of instability if not properly constrained.
• Applications range from advanced robotic impedance control and bacterial chemotaxis to astrophysical outflows and efficient information propagation in feedback networks.

Positive velocity feedback describes a class of feedback mechanisms in which the velocity (or related dynamic variable) of a physical, biological, or engineered system is fed back with a positive gain, thereby reinforcing ongoing motion or dynamic state. In various disciplines, positive velocity feedback can enhance system responsiveness, propagate information or alignment, or, when not properly constrained, destabilize passive behavior. Its mathematical and conceptual foundations manifest in domains from biological navigation and robotic control to cosmic gas outflows and collective particle dynamics.

1. Mathematical Foundations and General Mechanisms

Positive velocity feedback fundamentally involves the dynamic reinforcement of a system’s state based on the magnitude and direction of its velocity. Letting $v(t)$ denote the instantaneous velocity, a positive feedback loop adds a term proportional to $+K_v v(t)$ (or $v(t-\tau)$ for delayed feedback) to the governing equation of motion or control input, with $K_v > 0$ as the feedback gain. This contrasts with negative feedback, which introduces stabilizing or dissipative terms.

In many-faceted contexts, the mathematical realization may involve direct velocity terms (robotics, electronics), nonlinear dependence via motile probabilities (e.g., run-and-tumble microorganisms), or delayed velocity differences (delayed-feedback active matter). The sign and magnitude of the gain, as well as the architecture of nested feedback loops and memory, determine whether the velocity feedback augments responsiveness, produces a threshold for dynamic onset, or endangers stability and passivity.

2. Biological Navigation: Run-and-Tumble with Positive Feedback

In chemotactic navigation, as modeled for run-and-tumble microorganisms, positive velocity feedback arises when the internal state governing reorientation (tumble rate $\lambda(x)$) is itself modulated by recent motion up a sensory gradient. The drift-dynamics of the internal variable $x(t)$ are

$$\frac{dx}{dt} = -\frac{x - x_0}{t_M} + \frac{v_0 r}{t_E}$$

where $v_0$ is run speed, $r$ the projection of swimming direction along the gradient, $t_M$ the memory timescale, and $+K_v v(t)$0 the gradient feedback timescale. For $+K_v v(t)$1, positive feedback dominates, producing large excursions of $+K_v v(t)$2, and hence $+K_v v(t)$3, leading to amplification of runs in favorable directions.

Non-normal dynamics (non-orthogonal eigenvectors near the system's stable fixed point) and quadratic nonlinearities in the evolution equations enable the system to rectify fluctuations—resulting in the elongation of “uphill” runs and rapid truncation of “downhill” runs. In the high-gain regime, the average drift velocity $+K_v v(t)$4 saturates at half the maximal run speed ($+K_v v(t)$5), producing a molecular-scale ratchet that efficiently rectifies noise into directed ascent (Long et al., 2016).

3. Robotic Systems: Impedance Control with Positive Velocity Feedback

In advanced robot impedance control architectures, an inner-loop positive velocity feedback path substantially increases the closed-loop bandwidth of the torque response. The canonical structure consists of a PI-regulated torque loop, a velocity feedback block $+K_v v(t)$6 (or filtered variants), and an outer impedance controller. The equivalent plant dynamics become

$+K_v v(t)$7

where mechanical and electrical parameters are encoded in polynomials $+K_v v(t)$8. Stability is preserved only if $+K_v v(t)$9 does not exceed a limit set by the location of system zeros. Typically, $v(t-\tau)$0 with $v(t-\tau)$1. For the HyQ HAA joint, $v(t-\tau)$2 optimized bandwidth (from ≈80 Hz to ≈240 Hz) just shy of instability.

Passivity is not guaranteed in the presence of positive velocity feedback, as it can inject net energy into the system. While the full impedance loop may remain passive for certain parameter choices, exceeding critical $v(t-\tau)$3 (or increasing filter lag/step T) leads to passivity loss and potential instability. A thorough, stepwise design methodology ensures inner-loop performance gains do not compromise outer-loop impedance stability or passivity (Focchi et al., 2014).

4. Active Matter and Collective Dynamics: Delayed Feedback-Induced Propulsion

In systems of repulsively interacting Brownian particles with time-delayed feedback, positive velocity feedback manifests as propulsion and spontaneous velocity alignment without explicit steering rules. The feedback force derived from position over time-lag $v(t-\tau)$4,

$v(t-\tau)$5

with $v(t-\tau)$6 Gaussian in displacement, is effectively a delayed, repulsive “nudge.” For sufficiently strong feedback (threshold $v(t-\tau)$7), persistent particle velocities emerge, with alignment reinforced via memory and local steric effects. The positive definiteness of $v(t-\tau)$8 (feedback stiffness) mathematically encodes positive velocity feedback.

Large-scale alignment arises only if feedback-driven information can propagate: the distance traversed per delay must exceed interparticle separation. The resulting order–disorder transition, critical for emergent polar order, is absent in standard active Brownian models that rely on motility-induced phase separation (MIPS). The delayed feedback route leads to persistent alignment, transient clustering, and self-propulsion tunable via feedback parameters, distinguishing it from Vicsek or ABP models (Kopp et al., 2023).

5. Astrophysical Outflows: AGN-Driven Positive Radial-Velocity Feedback

In galaxy cluster evolution, active galactic nucleus (AGN) feedback generates large-scale, positive feedback in gas velocities via outflows of heated or mechanically ejected material. Thermal or kinetic feedback mechanisms deposit energy into the intracluster medium, producing bubbles with positive radial velocities $v(t-\tau)$9. These outflows expel gas to radii of $K_v > 0$0 times the virial radius, suppressing infall velocities and significantly altering gas and velocity power spectra (up to 60% suppression at $K_v > 0$1/Mpc, depending on feedback strength).

Jet (kinetic) prescriptions generate more collimated energy input, producing higher-velocity outflows at large radius relative to thermal models. As a function of cluster gas fraction deficit ($K_v > 0$2), outflow maximum velocities scale as $K_v > 0$3, with anisotropies in angular profiles and observable signatures in kSZ maps. Positive velocity feedback in this context leads to distinctive large-scale redistribution of baryonic matter and is a critical factor in accurately modeling cosmic velocity fields (Ondaro-Mallea et al., 2024).

Model	$K_v > 0$4	$K_v > 0$5	$K_v > 0$6
fgas+2σ	+0.03	0.45	4.8
fgas	0.00	0.80	6.1
fgas−2σ	−0.05	1.05	7.5
fgas−8σ	−0.15	1.25	9.8
Jet_fgas	0.00	0.70	5.4
Jet_fgas−4σ	−0.08	1.10	8.5

6. Information Propagation in Feedback Networks: Positive Information Velocity

In cascaded Gaussian (and more generally, additive noise) communication networks, feedback at each node enables the propagation of information at strictly positive velocity. Employing a linear minimum mean-squared error (MMSE) estimation scheme with feedback, each relay node transmits the innovation term, leading to recursive error reduction. The core recursion for the mean-squared error $K_v > 0$7 as a function of position $K_v > 0$8 and time $K_v > 0$9 is

$\lambda(x)$0

where $\lambda(x)$1 is channel signal-to-noise (per-symbol power constraint). For velocity $\lambda(x)$2, the error decays like $\lambda(x)$3, and the achievable information velocity (IV) is strictly positive for all arrival rates $\lambda(x)$4, the channel capacity. The positive velocity arises from the ability of feedback to reinforce the propagation of innovations along the cascade—a directly analogous mechanism to physical positive velocity feedback (Domanovitz et al., 2023).

7. Constraints, Instabilities, and Functional Considerations

While positive velocity feedback can dramatically increase system sensitivity, bandwidth, or alignment, unconstrained positive feedback tends to reduce effective damping and may violate passivity criteria. For engineered systems (robotics, electronics), explicit gain bounds must be enforced to prevent instability or net energy injection. In natural or active matter systems, positive feedback must be balanced by adaptation, negative feedback, or nonlinear rectification to avoid runaway behavior.

In summary, positive velocity feedback is a unifying principle that leverages the reinforcement of motion to achieve enhanced system-level outcomes. Its implications span the efficient rectification of stochasticity into drift (bacterial chemotaxis), expanded control bandwidth (robotics), emergent alignment (active matter), cosmological gas redistribution (AGN feedback), and information flow in networked systems. The utility and risks of positive velocity feedback are critically modulated by nonlinearities, gain structure, system architecture, and the presence of stabilizing countereffects.