To understand why the problem is difficult, it helps to unpack the hardware. A divert-and-attitude-control system, usually abbreviated DACS, uses solid-propellant thrusters to shove the interceptor sideways and to rotate its body. Unlike a missile with aerodynamic fins, a DACS-equipped kill vehicle can translate laterally even while its nose stays pointed where it was. That decoupling is a blessing for maneuverability and a curse for the seeker. Many exo-atmospheric interceptors carry their infrared seeker offset from the longitudinal axis of the body, an off-axis configuration that widens the usable field of regard and simplifies the mechanical layout. The price is a hard constraint: if the interceptor’s body pitches or yaws too far relative to the line of sight to the target, the target drifts out of the seeker’s field of view, and the engagement is effectively lost. Conventional proportional navigation, the workhorse guidance scheme dating back decades, commands the interceptor to rotate its velocity toward the target, and in doing so it can easily drive the body attitude out of the seeker’s cone.

The Beihang team’s answer is built around a quantity that guidance engineers have trusted for more than half a century: the zero-effort-miss, or ZEM. The ZEM is the predicted miss distance, the distance by which the interceptor would miss the target if both vehicles stopped maneuvering from that instant onward and simply coasted. It condenses the entire future of the engagement into a single vector that the guidance computer can watch in real time. If the ZEM shrinks toward zero, a collision is on track; if it grows or refuses to settle, the interceptor must act. Classical proportional navigation can be derived as the law that drives the ZEM to zero, and the closed-form solutions of pure and true proportional navigation worked out by earlier researchers remain foundational references in the field. What classical laws do not do, however, is promise anything about how fast the ZEM converges or how small it will be when time runs out.

That gap is precisely where the new work positions itself. The authors construct a generalized form of the zero-effort-miss suited to their interceptor’s dynamics and then design the guidance law so that the ZEM is guaranteed to converge, in finite time, into a small neighborhood that the designer chooses in advance. The mathematical machinery behind this guarantee is prescribed performance control, a technique introduced in the adaptive control literature by Bechlioulis and Rovithakis, in which a prescribed performance function wraps the tracking error in a decaying boundary. As long as the error stays inside that funnel, its transient behavior and its final bound are both known before the controller ever runs. Applied to guidance, the idea means the designer can specify, up front, the terminal miss-distance accuracy the interceptor must achieve and the shape of the convergence path, and the control law then enforces those bounds throughout the homing phase.

Prescribed performance ideas have appeared in guidance research before. Earlier studies used them to impose terminal line-of-sight angle constraints against maneuvering targets, to handle missile autopilot lag, and to build adaptive sliding-mode guidance laws. Robust intercept guidance with predesigned ZEM convergence has also been explored for capturing maneuvering targets. The distinctive contribution of the Beihang paper is the combination: a generalized ZEM formulation, a prescribed performance controller enforcing finite-time convergence of that ZEM within prescribed terminal bounds, and, critically, a design that holds the look angle fixed so the off-axis seeker never loses its lock. The fixed look angle is not an afterthought but a structural feature of the law. By keeping the angle between the interceptor’s body axis and the line of sight constant, the guidance commands are shaped so that the target remains comfortably inside the seeker’s field of view from the start of the engagement to the moment of closest approach.

The interplay between the two requirements is subtle. A DACS kill vehicle has two channels of actuation: divert thrusters that produce lateral acceleration and attitude thrusters that rotate the body. Spending divert authority on correcting the ZEM and attitude authority on managing the look angle would be the obvious division of labor, but the two channels are coupled through the engagement geometry, and both consume finite propellant. The guidance law must therefore thread a needle, driving the ZEM into its prescribed funnel fast enough to guarantee a small miss distance against a maneuvering target, while simultaneously respecting the seeker’s field-of-view limit that earlier work on strapdown and off-axis seekers has shown to be a first-class constraint rather than a nuisance. Studies of line-of-sight rate estimation for off-axis seekers on two-axis gimbals and of guidance strategies accounting for minimum field-of-view constraints have documented how easily naive commands violate the limit, particularly when the initial heading error is large or the target maneuvers late.

Maneuvering targets are the stress test for any homing law. In an exo-atmospheric scenario, the target may be a ballistic warhead executing evasive lateral pulses in the last seconds of flight, and every meter of miss distance matters because the interceptor’s lethality radius is finite. The ZEM formulation is attractive here precisely because target acceleration enters the prediction directly: an unanticipated target maneuver shows up immediately as a jump in the ZEM, giving the guidance law an unambiguous error signal to cancel. By wrapping the generalized ZEM in a prescribed performance function, the Beihang design converts that error signal into commands with a guaranteed convergence envelope, so that even under maneuvering, the miss distance is forced into the small terminal neighborhood the designer prescribed, in finite time rather than merely asymptotically. Asymptotic convergence, the standard promise of many classical controllers, is a luxury an interceptor does not have; the engagement ends at a definite instant, so the guarantee must hold before that instant arrives.

The authors validate the approach through extensive numerical simulations, and they state that all mathematical models and simulation parameters are detailed in the manuscript, facilitating direct implementation in MATLAB/Simulink. That level of reproducibility matters in a field where claims of terminal accuracy are easy to make and hard to check. The simulation campaign, illustrated across the paper’s figures, exercises the guidance law in the scenarios that matter: engagements of maneuvering targets by a DACS-equipped, off-axis-seeker interceptor, with the ZEM trajectory, the look angle history, and the resulting miss distance all under scrutiny. The reported behavior supports the central claim, that the ZEM converges within the prescribed bounds while the seeker maintains continuous lock throughout the engagement.

The broader context for this work is the long-running evolution of interceptor guidance. The literature spans THAAD-like high-altitude theater missile defense analyses, optimal guidance-to-collision laws for accelerating exo-atmospheric interceptors, energy-effective guidance-to-collision strategies, blended control strategies for high-speed targets at high altitude, and collision-control guidance based on speed control for kill vehicles. Sliding-mode precursor interceptor guidance, solid-propulsion attitude control for exo-atmospheric kill vehicles, switching logic for DACS, and DACS missile controllers that explicitly consider the seeker’s field-of-view limit have all chipped away at pieces of the problem. Unified formulations of optimal guidance-to-collision laws for accelerating and decelerating targets and nonlinear optimal impact-angle-constrained guidance with large initial heading error round out a rich toolbox. What most of these methods share is a focus on the collision condition itself; the Beihang law instead focuses on sculpting the convergence of the predicted miss distance, with the seeker constraint folded into the design from the first equation.

For the field, the significance is twofold. Practically, the law offers designers of off-axis-seeker interceptors a way to guarantee terminal accuracy without gambling on the seeker’s field of view, which could simplify integration of offset sensor architectures on future kill vehicles. Methodologically, it demonstrates that prescribed performance control, born in the adaptive control of strict-feedback systems, extends cleanly to the generalized ZEM dynamics of a space intercept, providing finite-time, bounded-error convergence guarantees in a domain where such guarantees are rare and valuable. The work, received in October 2025 and published on 14 October 2026 after revision, was communicated by Zongyi Guo, and the authors acknowledge Prof. Wanchun Chen’s guidance throughout the research. As missile defense engagements push to higher altitudes, higher closing speeds, and more agile evaders, guidance laws that make explicit, provable promises about the final miss distance, while respecting the physical limits of the sensors that must see the target through to the end, are likely to shape the next generation of kill vehicle design.

Subject of Research: Prescribed-performance homing guidance for exo-atmospheric interceptors with off-axis seekers and divert-and-attitude-control systems

Article Title: Generalized Intercept Guidance Law with Prescribed Zero-Effort-Miss Convergence for Off-Axis Seeker-Based Interceptors

Article References: Mushtaq, M., Yu, W., & Chen, W. (2026). Generalized Intercept Guidance Law with Prescribed Zero-Effort-Miss Convergence for Off-Axis Seeker-Based Interceptors. International Journal of Aeronautical and Space Sciences. https://doi.org/10.1007/s42405-026-01307-3

Image Credits: AI Generated

DOI: 10.1007/s42405-026-01307-3

Keywords: guidance law, zero-effort-miss, off-axis seeker, exo-atmospheric interceptor, prescribed performance control, DACS, maneuvering target, homing guidance, miss distance, Beihang University, finite-time convergence, field of view

The core challenge the researchers tackled is one that has haunted robotics engineers for decades. A robotic manipulator is a strongly coupled, nonlinear system: moving one joint changes the forces acting on every other joint, and the equations governing its motion involve inertia, Coriolis and centrifugal effects, gravity, and friction that all shift as the arm moves through space. When the actual mass or inertia of a link differs from the values in the controller’s internal model, when Coulomb and viscous friction resist motion in ways that are hard to characterize, or when external torques push on the arm, tracking accuracy degrades and stability can be compromised. The team, led by Xiaole Ma and corresponding author Chenghu Jing, built their approach on a rigorous mathematical model of the UR10 that explicitly accounts for these perturbations, lumping all the unknown dynamics into a single disturbance term that their controller must confront head-on.

Their solution rests on a division of labor between two complementary technologies. The first is a radial basis function neural network, a lightweight learning architecture prized for its simple structure and its ability to approximate any continuous nonlinear function within a defined region. Placed in the feedforward path of the controller, the network observes the robot’s state in real time and learns to predict the lumped unknown dynamics, from friction quirks to inertia mismatches. By absorbing the dominant nonlinearities before they can corrupt the motion, the network dramatically shrinks the size of the residual disturbance that the robust part of the controller must handle. This matters because the size of that residual directly determines how aggressive, and how energy-hungry, the robust control term needs to be.

That robust term is where the super-twisting algorithm enters the picture. Sliding-mode control is a classic robust technique: it drives the system’s tracking error onto a carefully chosen sliding surface and then uses aggressive switching to hold it there, making the system immune to a broad class of disturbances. The trouble is that conventional sliding-mode control relies on discontinuous switching, which produces the infamous phenomenon of chattering, a high-frequency oscillation in the control torque that excites unmodeled vibrations, wastes energy, and accelerates wear on actuators and gears. The super-twisting algorithm, a second-order sliding-mode technique, solves this problem elegantly. Instead of switching the torque itself, it applies a continuous combination of a term proportional to the square root of the sliding variable’s magnitude and an integral term, achieving the same finite-time convergence and disturbance rejection while producing a smooth, continuous control signal.

The Chinese team’s contribution is a carefully engineered synthesis of these two elements, backed by a rigorous stability proof. They defined a linear sliding surface based on the tracking error and its derivative, then constructed a control law with three modules: a nominal dynamic compensation term built from the robot’s known model, the neural network feedforward term for online approximation of unknown dynamics, and the super-twisting robust term to suppress residual disturbances. To keep the neural network’s weights from drifting to unbounded values, a common failure mode of adaptive schemes, they introduced a projection-based adaptive law that mathematically guarantees the weight estimates never exceed a preset safety boundary. Using Lyapunov stability theory, the researchers proved that both the sliding variable and the tracking error converge in finite time, with explicit gain conditions relating the super-twisting parameters to the disturbance bound.

The proof framework itself represents a departure from conventional practice. Rather than using the adaptive law to exactly cancel uncertainty terms, which can lead to gain drift and complicated derivations, the team adopted a decoupled design in which the neural network physically reduces the total disturbance bound. This allows the super-twisting gains to be substantially relaxed, cutting energy consumption and further suppressing chattering while preserving the finite-time stability guarantee. It is a design philosophy that treats the neural network and the robust controller as partners rather than competitors, each doing the job it is best suited for.

To test the approach, the researchers ran simulations on a full six-degree-of-freedom UR10 model using experimentally identified nominal parameters. They deliberately made the task hard: the masses of joints two and three were increased by twenty percent, centers of mass were shifted, inertias were perturbed by five percent, and a smooth sigmoidal friction model captured Coulomb and viscous effects in every joint. A sinusoidal external disturbance of 0.5 newton-meters was applied to all joints, and a fixed-gain study showed the method maintained tracking errors below 0.005 radians even when disturbances reached 2.0 newton-meters. The robot was commanded to move from one pose to another along a fifth-order polynomial trajectory over ten seconds, with a control period of one millisecond.

The results were emphatic. Under pure neural network control, lacking any robust term, the steady-state errors of the wrist joints ballooned to as much as 0.82 radians, with maximum errors exceeding one radian, an unacceptable failure. Conventional sliding-mode control fared far better, reducing steady-state errors to the thousandth-of-a-radian range, but the proposed composite controller cut them further still: joints four and five achieved steady-state errors of 0.0005 and 0.0017 radians, improvements of more than fifty percent and seventy-eight percent over pure sliding-mode control. All joints except one achieved steady-state errors better than one-thousandth of a radian. Measured by the integral of time-weighted absolute error, a standard metric combining response speed and steady-state accuracy, the composite controller outperformed pure sliding-mode control by 86.1 percent and pure neural network control by 99.7 percent. In Cartesian space, the end effector’s steady-state error shrank to 0.0002 meters, compared with 0.0007 meters for sliding-mode control and a disastrous 0.1031 meters for the network alone.

Just as important as the accuracy is the character of the control signals. The composite controller produced continuous, smooth torques free of the high-frequency chattering that traditional sliding-mode schemes generate, with peak torques around 62 newton-meters at joint two, comfortably below actuator saturation limits. The neural network’s weights, rather than growing without bound as they did under pure network control, stabilized at a low value after a brief transient, confirming that the super-twisting term was shouldering the main disturbance rejection burden while the network handled fine compensation. The method also proved its generality on continuous sinusoidal trajectories simulating industrial operation, with all joint root-mean-square errors below 0.003 radians. The authors note that future work will focus on experimental validation on a physical UR10 platform, adaptive optimization of the network structure, and integration of iterative learning with neural sliding-mode control. If hardware trials confirm the simulation results, this hybrid of learning and robust control could find a home in precision assembly, medical assistance, and any application where a robot arm must move exactly where it is told, even when the world pushes back.

Subject of Research: Composite neural network and super-twisting sliding-mode control for high-precision trajectory tracking of the UR10 robotic manipulator

Article Title: Trajectory-tracking control of the UR10 manipulator based on radial basis function neural network and super-twisting sliding mode

Article References: Ma, X., Jing, C., Zhang, K., Chen, C., & Wang, Y. (2026). Trajectory-tracking control of the UR10 manipulator based on radial basis function neural network and super-twisting sliding mode. Mechanical Sciences, 17(2), 759-767. https://doi.org/10.5194/ms-17-759-2026

Image Credits: AI Generated

DOI: 10.5194/ms-17-759-2026

Keywords: robotics, UR10 manipulator, trajectory tracking, sliding-mode control, super-twisting algorithm, radial basis function neural network, chattering suppression, adaptive control, Lyapunov stability, finite-time convergence, collaborative robots, nonlinear friction

The approach centers on an adaptive finite-time disturbance observer, a software mechanism that watches the satellite’s rotational behavior, infers the hidden torques being injected by faults and environmental disturbances, and then cancels them through compensation in the control command. Rather than treating every anomaly the same way, the scheme explicitly models the three dominant actuator failure modes encountered during long-duration on-orbit operation: loss of effectiveness, in which an actuator delivers only a fraction of the commanded torque; bias faults, in which a constant unwanted torque is added on top of the command; and jamming, in which an actuator seizes and stops responding altogether. By bundling all of these effects, together with external disturbances such as gravity-gradient and aerodynamic torques, into a single lumped disturbance signal, the observer reduces a complicated multi-fault diagnosis problem to a well-defined estimation task that can be solved in real time onboard the spacecraft.

What distinguishes the new design from earlier disturbance observers is the way it balances three competing demands that have long troubled control engineers: speed of estimation, smoothness of the resulting control action, and physical plausibility of the estimates. The first mechanism is an error-norm-based adaptive gain. Classical high-gain observers estimate disturbances quickly, but the aggressive gain amplifies sensor noise and produces jittery control commands that waste fuel and fatigue actuators. Low-gain designs are smooth but sluggish, leaving the satellite pointing erratically for extended periods after a fault occurs. The adaptive gain interpolates between these extremes: when the estimation error is large, as it is immediately after a fault, the gain rises to accelerate convergence; as the error shrinks, the gain relaxes, preserving smoothness. In the reported simulations, this adaptive gain alone reduced the post-fault control total variation by 58 percent compared with a fixed high-gain design that achieved the same pointing accuracy.

The second mechanism is a double-power injection law, a nonlinear feedback term built from signed power functions of the observer’s auxiliary variable. One power exponent, chosen between zero and one, dominates close to the origin and eliminates the slow asymptotic creep that plagues linear observers near convergence; a higher-order power exponent, greater than one, dominates far from the origin and rapidly pulls large initial errors back into range. The combined effect drives the observer’s auxiliary variable into a small residual neighborhood of the origin in finite time rather than merely as time approaches infinity, and the resulting estimation error is proven to be uniformly ultimately bounded. Finite-time convergence matters operationally because every second the observer spends hunting for a fault is a second the satellite is pointing in the wrong direction—a serious concern for Earth-observation missions, laser communication links, and formation-flying constellations that demand tight angular tolerances.

The third mechanism is a projection operator, a mathematical guardrail that confines the disturbance estimate to a predefined physically admissible set. Without such a constraint, an observer reacting to noisy gyroscope measurements or an unmodeled transient can generate absurd estimates—an apparent disturbance torque far beyond anything the environment or a failed actuator could produce—and the compensating command built on that estimate would be equally absurd, potentially destabilizing the spacecraft. The projection operator enforces bounds derived from the known limits of the actuators and the expected disturbance environment, ensuring that even under pathological conditions the compensation signal remains sane. Together, the three mechanisms form a layered defense: adaptive gain governs how hard the observer works, the double-power law governs how quickly it converges, and the projection operator governs what answers it is permitted to give.

Beyond the observer itself, the paper contributes a theoretical analysis of the estimation–compensation loop that arises when the dominant fault is a loss of actuator efficiency. This loop is subtle: the observer estimates the lumped disturbance, the controller compensates for it, the compensation changes the actual torque, and the changed torque feeds back into the disturbance estimate. The authors show that for every axis whose actuator retains non-zero efficiency, this loop is contractive—it draws estimates toward a single point rather than amplifying errors—and that the unique fixed point of the loop recovers the nominal closed-loop dynamics exactly. In plain terms, once the observer converges, a satellite with a weakened reaction wheel behaves, from the perspective of its attitude control loop, as if the wheel were healthy. This kind of exact recovery guarantee is rare in fault-tolerant control, where results more often promise bounded error rather than restoration of nominal behavior.

The validation campaign is unusually thorough for this class of study. The authors ran ablation simulations in which each of the three observer mechanisms was individually removed, confirming that each contributes measurably to the overall performance. They tested multiple fault types and magnitudes, spanning gradual efficiency losses, sudden biases, and complete jams, and they stressed the observer with external disturbance amplitudes up to fifty times the nominal environmental level. They also injected realistic gyroscope measurement noise into the loop to verify that the adaptive gain’s noise-suppression advantage holds under sensing conditions close to those of a real spacecraft. Across all of these scenarios, the scheme maintained its near-complete recovery of pointing accuracy, with the 0.013-degree steady-state error achieved even as disturbance levels varied by orders of magnitude.

The practical implications extend across the growing orbital economy. Modern small satellites have transformed access to space but often fly with minimal redundancy, so a single degrading reaction wheel can end a mission that cost millions of dollars to build and launch. Mega-constellations, Earth-observation platforms, and science missions all depend on attitude control systems that must survive years of wear, radiation damage, and lubricant degradation. An observer-based scheme like this one requires no additional hardware, no redundancy in the actuator set, and no ground-in-the-loop diagnosis: it runs as software, detects the aggregate effect of a fault through its dynamic signature, and compensates autonomously within the finite convergence time of the observer. For operators, that could mean the difference between degrading a mission’s pointing budget and silently absorbing a hardware failure that no one on the ground even notices until telemetry is reviewed.

The work also fits into a broader resurgence of disturbance-observer-based control, a lineage running from Luenberger observers through active disturbance rejection control and modern fixed-time stabilization theory. By proving contraction of the estimation–compensation loop, establishing uniform ultimate boundedness of the estimation error, and demonstrating quantitatively that adaptation buys smoothness at no cost in accuracy, the study offers a template that engineers in adjacent fields—reusable launch vehicles, flexible spacecraft with vibrating appendages, and autonomous drones facing wind gusts—may adapt to their own fault-tolerant control problems. The research was supported by the Aerospace-Power Fund Project of the China Space Foundation under Grant No. KDJJ2025040121. As satellites multiply in orbit and missions grow more ambitious, control laws that let a wounded spacecraft heal its own pointing, silently and in finite time, may become one of the quiet workhorses of the space age.

Subject of Research: Adaptive finite-time disturbance observer-based fault-tolerant attitude control for satellites experiencing actuator failures

Article Title: Adaptive Finite-Time Disturbance Observer-Based Fault-Tolerant Control for Satellite Attitude Systems with Actuator Failures

Article References: Yang, Q., Gu, J., Wang, Q., Cui, P., Lu, B., & Lu, H. (2026). Adaptive Finite-Time Disturbance Observer-Based Fault-Tolerant Control for Satellite Attitude Systems with Actuator Failures. International Journal of Aeronautical and Space Sciences. https://doi.org/10.1007/s42405-026-01287-4

Image Credits: AI Generated

DOI: 10.1007/s42405-026-01287-4

Keywords: satellite attitude control, fault-tolerant control, actuator failures, disturbance observer, adaptive gain, finite-time convergence, spacecraft control, pointing accuracy, reaction wheels, projection operator, control systems, small satellites