Throwing a Tight Spiral American Football by a Humanoid Robot

Zaid Mahboob; Bowen Weng

cs.ROAug 17, 2026

Throwing a Tight Spiral American Football by a Humanoid Robot

Authors: Zaid Mahboob, Bowen Weng

Organizations: Department of Computer Science, Iowa State University

Abstract

Accurate throwing of the American football requires precise regulation of release conditions, where coupled linear and angular momentum determine flight stability and targeting accuracy. While prior work on robotic object throwing has largely focused on generating dynamically feasible release velocities using open-gripper paradigms, explicit control of spin injection at detachment remains underexplored, particularly for aerodynamically anisotropic objects like the American football. In this paper, we present the spin-stabilized controlled tight spiral throw of an American football by a humanoid robot. Achieving this requires (i) accurately reaching the desired coupled momentum, which often involves high degrees-of-freedom (DoF) movements completed within approximately half a second, and (ii) managing the complex transient contact dynamics that arise during the sub-100-millisecond release phase, when the football is effectively underactuated as it moves partially across the fingers. To this end, we develop a coupled whole-body control strategy where the lower body is performing informed stabilization while the upper body is further divided into two phases with (i) a throw phase accelerating the football to a target state through trajectory optimization and tracking, and (ii) a follow-through phase utilizing model predictive control to actively control the wrist and remaining in-contact fingers. The proposed framework is empirically validated on a 29-DoF Unitree G1 humanoid equipped with a 7-DoF Dex3-1 three-fingered gripper. The thrown American football reaches up to 93.6% spin efficiency and a 0.286 radians linear-velocity-to-nose-alignment (nose-angle) error (where an "ideal" tight spiral corresponds to 100 % spin efficiency and 0 radians nose-angle error) at up to a 5.35 m/s linear velocity and an angular velocity of 14.5 rad/s.

Figures & tables

Fig. 1: Demos of a humanoid robot (Unitree G1 with Dex3-1) throwing a tight spiral American football. The controller has two sequenced phases: the throw phase generates the coupled momentum, while the follow-through phase uses model predictive control (MPC) to regulate the release dynamics and stabilize the tight spiral. Without follow-through (top left), the ball tumbles due to an unstable release; with follow-through, the ball has a stable tight spiral (top right).

Fig. 2: Football rigid-body state and kinematic quantities used in the tight spiral Definition 1 . The football state is defined as

\mathbf{x}_{\text{b}}=(\mathbf{p},R,\mathbf{v},\boldsymbol{\omega})

, where

\mathbf{p},\mathbf{v},\boldsymbol{\omega}\in\mathbb{R}^{3}

represent the center-of-mass position, linear velocity, and angular velocity, respectively, and

R\in\text{SO}(3)

is the football orientation matrix mapping vectors from the body frame to the world frame. The football longitudinal body-frame axis is defined as

\mathbf{e}_{x}=[1,0,0]^{\top}

, and the corresponding world-frame nose direction is given by

\hat{\mathbf{n}}=R\mathbf{e}_{x}

Fig. 3: The proposed whole-body control architecture along with the estimation solutions for the humanoid football throwing of a tight spiral throw.

Fig. 4: Illustration of the contact-aware football dynamics model, where $\mathbf{v}_{\text{surf}}$ , $\mathbf{v}_{\text{pad}}$ , and orange dots denote the contact point velocities of the football surface, the fingertip pad, and the sampled contact points, respectively.

Sym.	Value	Sym.	Value
Throw Phase
$\mathbf{w}_{\text{run}}$	$[10^{3},0.01,0.05,10^{3}]$	$T$	$0.60$ s
$\mathbf{w}_{\text{term}}$	$[10^{4},2{\times}10^{4},2{\times}10^{4}]$	$t_{a}$	$0.55$ s
$\mathbf{v}^{*}$	$[3,0,0]$ m/s	$\boldsymbol{\omega}^{*}$	$[-6,0,0]$ rad/s
$\text{w}_{\text{align}}$	$10^{6}$
Follow-Through Phase

TABLE I: Controller Parameters

Fig. 5: Temporal evolution of spin efficiency

\eta_{\text{spin}}

(left) and nose-velocity angle

\alpha_{\text{nose}}

(right) throughout multiple trials. The blue shaded band represents the distribution of multiple attempts of the throw phase using the presented method in Section III-A . At the point of release, follow-through ensures a clean finger separation, preserving high spin efficiency (shown by the green colored line). Conversely, without follow-through, uncontrolled release induces a pronounced wobble, characterized by decaying spin efficiency and progressive misalignment (shown in red). Each trace terminates upon complete finger separation and is held constant thereafter for visual comparison.

Fig. 6: Joint-level torque impulse distribution for the throw and follow-through phases. Each pie chart shows the normalized time integral of joint torque, with the top three contributing joints highlighted. The throw phase is dominated by shoulder roll, waist pitch, and shoulder pitch, while the follow-through phase is dominated by waist yaw, waist roll, and waist pitch.

Robotic Manipulation

Robotic manipulation studies how robots perceive, plan, and control physical interactions to grasp, move, assemble, and use objects.

See all 1,685 papers

Oct 7, 2026

Robotic Boomerang Throwing via Model-Based Release Design

Throwing objects that generate aerodynamic lift can greatly extend robot throwing beyond ballistic flight. A returning boomerang is a challenging example because its flight depends strongly on the release velocity, attitude, and spin, while robotic manipulators cannot readily reproduce the rapid motions used in human throwing. We present a model-based framework for robotic boomerang throwing centered on the release state. We identify the boomerang flight dynamics in stages to predict how flight changes across design variations. To systematically design the robot throwing motion, we screen candidate parameters according to how strongly and consistently they control release spin under uncertain contact conditions. These models are then used to design the throwing motion and boomerang for a 6-DoF manipulator with limited joint speeds. To our knowledge, this is the first robotic manipulator to generate a returning boomerang flight. In the demonstrated returning trial, the boomerang is released at 51 rad/s (8.1 rev/s), reaches 2.03 m from the robot base, and returns to touch down 0.31 m from the base. The successful release differs significantly from the measured human throws, showing that a robot need not imitate human throwing motion to achieve a returning flight. The project page is available at https://robot-boomerang.github.io

Yang Liu, Colin Jones, Aude Billard

Learning Algorithms and Systems Lab (LASA), EPFL, Switzerland · Automatic Control Lab, EPFL, Switzerland

Robotic Manipulation Dynamic Robotic Manipulation Robot Motion Generation

Ask PDF

Jun 1, 2026

FlipItRight: Stable Pose-Targeted Throw-Flip Across Diverse Objects

We propose FlipItRight, a framework for stable planar pose-targeted throw-flip with a high-DoF manipulator. The task is decomposed into an object-level planner, which generates candidate release states satisfying the desired landing pose, and a robot-level planner, which evaluates executability and constructs a feasible swing motion. Treating the release state as an explicit intermediate representation enables principled candidate filtering, adaptive selection of release and pre-swing configurations, and structured near-release motion design -- in particular, approximately constant end-effector velocities during the final swing phase to improve robustness to release-timing uncertainty. We validate on a real platform across objects of varying shape, size, and mass, achieving a 90% success rate across 120 trials. Ablation studies confirm that each design choice contributes to throwing performance, and the framework requires no prior data or learned model, enabling direct deployment on new objects and targets without environment-specific calibration or data collection.

Axel Dawne, Shinkyu Park

King Abdullah University of Science and Technology (KAUST), Saudi Arabia

Non-Prehensile Manipulation Robot Motion Generation Robotic Manipulation Robot Motion Planning

Ask PDF

Sep 1, 2026

Non-Prehensile Throwing: A Reinforcement Learning Perspective

Robotic throwing enables fast object transport and extends a robot's reachable workspace beyond traditional pick-and-place. While prehensile (grasp-based) throwing works well for graspable items, non-prehensile (grasp-free) throwing is better suited for large, heavy, and/or deformable objects. Existing approaches rely on model-based optimization with simplified contact models (e.g., dynamic grasping) and low-dimensional trajectory parameterizations, which limit solution quality and reachable workspace. We propose a reinforcement learning approach that additionally leverages sliding and rolling contact modes and directly optimizes joint-space trajectories without analytical contact models or custom parameterizations. The Markov Decision Process (MDP) is formulated as a dynamical system that evolves the robot's joint state conditioned on the throwing target, object model, and initial configuration. Joint-jerk trajectories are planned offline at a low control rate and upsampled into smooth, high-rate velocity commands for deployment. For sim-to-real transfer, we minimize the robot-dynamics gap through minimum-jerk system identification and train uncertainty-aware policies to mitigate object-modeling errors, particularly sensitivity to dynamic friction. In simulation, the policy achieves 99% success across thousands of configurations and generalizes to unseen objects. Sensitivity analysis shows robustness to mass uncertainty but high sensitivity to dynamic friction, consistent with the sliding-based release mechanism. Deployed zero-shot on a UR5e operating near its physical limits (5 m/s end-effector velocity), our method throws diverse objects including heavy (790 g) and large (20x20x28 cm) items to targets up to 350 cm distance or 180 cm elevation, achieving a 97% real-world success rate.

Abdullah Mustafa, Ryo Hanai, Ixchel G. Ramirez-Alpizar +4

National Institute of Advanced Industrial Science and Technology (AIST), Japan · Graduate School of Fundamental Science and Engineering, Waseda University, Tokyo 169-8555, Japan

RL for Robotics Non-Prehensile Manipulation Contact-Rich Robotic Manipulation Non-Prehensile Robotic Manipulation

Ask PDF

Abstract

Robotic Manipulation

Explore similar work

Robotic Boomerang Throwing via Model-Based Release Design

FlipItRight: Stable Pose-Targeted Throw-Flip Across Diverse Objects

Non-Prehensile Throwing: A Reinforcement Learning Perspective